uniteTOlove.ca

Surveillance Pricing in Toronto Groceries: A Shadow Report on Council Item 2026.EX33.33

Publish word given 2026-08-06 — not yet linked into site navigation. This is a shadow report on Toronto Council item 2026.EX33.33 ("surveillance pricing" in grocery retail), prepared and approved as a PDF and sent, on 2026-08-05, to the City staff assigned the Q1 2027 research directive. This page is a web render of that same approved content. It is reachable at this direct URL but not yet linked from any nav menu or footer, and not yet announced.
v1.2 Shadow report on Toronto Council item 2026.EX33.33, "Making Grocery Prices Fair: Banning Surveillance Pricing in Toronto." Source file: this library's internal records. v1.2, August 5, 2026 — one material update over v1.1: the final recorded Council vote (Adopt Item — Carried, 26–0, July 30, 2026) incorporated. Independent research draft, not a legal opinion and not a City of Toronto document.

v1.2 (August 5, 2026): one material update only — the final recorded vote on item 2026.EX33.33, posted to the City's agenda-item page since v1.1 closed, is now incorporated (Executive Summary; §3.1): Adopt Item — Carried, recorded vote 26–0, July 30, 2026. No other content changed; v1.1 is preserved beside this file as its snapshot.

Integration pass, August 1, 2026 — v1.1, the filing candidate: v1.0 plus the full disposition of a second-family outside review (Kimi k3-256k, per the a recorded standing decision symmetric review rule — verdict SHIP-WITH-FIXES, 19 findings, every one adjudicated and applied; Appendix I.5). Built in three same-day passes (nine parallel primary-source verification lanes in total) over the project's verified spine (sonnet-base/one of this library's internal records) and the 18-submission multi-AI capture harvest. All prior gates are closed: 45 claims register items verified, 16 tool claims rejected as fabricated or misattributed, zero unverified-commodity marks remaining in the body. Supersedes one of this library's internal records, one of this library's internal records and one of this library's internal records (all kept as historical records of the first two passes). Purpose, by editorial decision: a shadow report paralleling — and intended to pre-empt — the staff report Toronto Council directed for Q1 2027.

Verification status, at a glance. Three classes of claim appear in this report, and the reader should be able to tell them apart without hunting:

  1. Independently verified. Everything carried over from the master briefing (which was itself independently web-verified on July 31, 2026) plus everything the August 1 verification pass confirmed against live primary sources. The August 1 pass verified, among other things: the City of Toronto Act's actual ss. 6–8 and 11 text (closing the report's single biggest prior gap); Manitoba Bill 49's proclamation — in force July 1, 2026, a material correction to the master briefing; Ontario Bill 104's exact status; Libman v. The Queen, [1985] 2 S.C.R. 178 and its modern successors; Clearview AI Inc. v. British Columbia (Information and Privacy Commissioner), 2026 BCCA 67; Councillor Dianne Saxe's on-record quote; Bill C-36's actual definitional text; Bill C-226's second-reading vote; China's PIPL Article 24; and the true character of N.Y. Gen. Bus. Law §§ 340-b and 349-a. A same-day second pass verified 22 further items — Toronto's own MM24.42/MM39.27 record, the Croplife and Eng precedents, the Colorado veto letter, Dubé & Misra's JPE study, the Maryland/New Jersey/Connecticut/federal bill-number conflicts, and more. A third pass closed the final gates: the Retail Council of Canada's on-record letter filed on item 2026.EX33.33 itself, Connecticut's enrolled Public Act 26-64 (correcting its effective date to October 1, 2026), and the verified conflict-doctrine base for the s. 11 analysis in Chapter 3.5A. Full claims register: Appendix I ("Integration record").
  2. The marking law. During drafting, material from the multi-AI capture wave carried inline [UNVERIFIED-COMMODITY: <tool>] marks until verification touched it. As of v1.0, no marked material remains in the body — every borrowed claim has been verified against a primary source, corrected, or removed (Appendix I records all three outcomes). The law remains in force for any future edit: nothing new from a commodity tool enters this document unmarked, and nothing marked may be treated as settled.
  3. Checked and rejected. Claims from the wave that verification disproved are not silently dropped; they are recorded in Appendix I, because knowing what the commodity field got wrong is part of this project's evidentiary margin.

Status note: this report answers a live, moving target. Several of the laws and bills discussed below were mid-process as of August 1, 2026 (bills pending signature, consultations not yet reported out, appeals pending). Where status could change quickly, that is flagged explicitly rather than stated as settled.

Executive Summary

On July 30, 2026 — at its July 29–31 meeting, the last of the term before the municipal election — Toronto City Council adopted item 2026.EX33.33 unanimously, by a recorded vote of 26–0 (0 absent) per the City's posted decision record (see 3.1) — "Making Grocery Prices Fair: Banning Surveillance Pricing in Toronto," sponsored by Mayor Olivia Chow and Councillor Alejandra Bravo. The item does not itself ban anything. It directs City staff — the City Manager, City Solicitor, and Executive Director of Municipal Licensing and Standards — to report back to the newly elected term of Council in the first quarter of 2027 with (1) all possible mechanisms to ban or regulate surveillance pricing on groceries, (2) a jurisdictional scan running from mandatory plain-language disclosure up to full prohibition, consumer opt-out rights, and electronic-shelf-label transparency, (3) a request that Ontario ban the practice and strengthen the Consumer Protection Act, (4) collaboration with the federal Food Security Strategy and AI Strategy, and requests for advice from (5) Innovation, Science and Economic Development Canada (ISED) on PIPEDA, (6) the Competition Bureau of Canada, and (7) the Office of the Privacy Commissioner of Canada, plus (8) forwarding the item to the Association of Municipalities of Ontario and the Federation of Canadian Municipalities.

What surveillance pricing is. "Surveillance pricing" — the Competition Bureau of Canada's own June 2025 discussion paper treats "personalized pricing" and "surveillance pricing" as synonyms — means using data about a specific individual (browsing history, location, device, inferred income, purchase timing) to set that individual's price, as distinct from ordinary "dynamic pricing" keyed to market conditions (demand, time, inventory) that applies uniformly to everyone. The U.S. Federal Trade Commission's January 2025 "Issue Spotlight" deliberately declines to adopt one legal definition, instead describing the phenomenon and its enabling technologies (data brokers, AI/ML pricing engines, e-commerce and in-store infrastructure including electronic shelf labels).

Is it actually happening in Toronto or Canada today? This is the report's single most important finding, and it cuts against the alarm implicit in the council item's framing: no investigation, regulator, or journalist has documented a Canadian grocer using individualized, personal-data-driven pricing on identical items today. Loblaw, Sobeys, and Metro have all told Canadian journalists on the record that they do not and will not use algorithmic personalization, and the electronic-shelf-label vendor Pricer has said its systems require human-approved, store-wide price changes, not per-customer ones. The Competition Bureau's own June 2025–January 2026 consultation on algorithmic pricing found no concrete Canadian examples, though it has since opened a broader 2026–27 review of the food supply chain (loyalty programs, algorithmic pricing, "shrinkflation") precisely because the capability is spreading even if confirmed practice is not. By contrast, the underlying technology and a real record of misuse are well documented in the United States: Groundwork Collaborative and Consumer Reports caught Instacart running undisclosed AI price experiments (the "Eversight" tool) producing up to 23% price differences on identical simultaneous orders, prompting an FTC probe and Instacart's discontinuation of the practice in December 2025; Consumer Reports separately documented Kroger's "income predictor" shaping discount targeting; and a peer-reviewed study (Aparicio, Metzman & Rigobon, Quantitative Marketing and Economics, 2024) found major U.S. online grocers already personalize prices by delivery ZIP code. This finding held up under a platform-by-platform check (Chapter 1.4): DoorDash, Uber Eats, Amazon/Whole Foods, and Walmart Canada all operate grocery delivery in Toronto with no Canada-specific surveillance-pricing finding against any of them, and the same is true of every Canadian-owned platform checked — Loblaw's PC Express/PC Optimum, Sobeys' Voilà, Metro.ca, and Longo's/Grocery Gateway. Toronto's motion is therefore precautionary and capability-driven, not a response to a proven local harm — a distinction the report below preserves throughout, and one the City's own eventual 2027 staff report will need to state plainly.

Does Toronto actually have the legal authority the motion assumes? This is contested, and the report takes both sides seriously (Chapter 3). The origin letter asserts, without citing a specific statutory section, that "the City of Toronto Act provides authority to regulate for the purpose of consumer protection and the economic well-being of Torontonians." This integration pass located and verified the actual statutory anchors the letter omits: the City of Toronto Act, 2006 expressly authorizes by-laws respecting "Protection of persons and property, including consumer protection" (s. 8(2), para. 8), the "Economic, social and environmental well-being of the City" (para. 5), and "Business licensing" (para. 11), read under s. 6(1)'s broad-interpretation command — so the claimed hook is real, not invented. The honest counterweight is not a missing power but s. 11: a City by-law is without effect to the extent it conflicts with, or frustrates the purpose of, any provincial or federal Act — the precise ground on which a city-wide prohibition overlapping the Consumer Protection Act and Competition Act would be attacked. Chapter 3.5A runs that attack against the governing conflict doctrine (Spraytech, Rothmans, Croplife, Canadian Western Bank — all verified against the primary judgments) and concludes it is survivable: dual compliance is possible, and neither the Consumer Protection Act nor the Competition Act — which repealed its price-discrimination offence in 2009 — contains an expressed purpose a stricter municipal rule would frustrate. Legal academics quoted in Canadian Press/CHCH coverage — Alberto Salazar (Carleton University), Dan Cohen (Queen's University), and Jake Okechukwu Effoduh (Toronto Metropolitan University) — are skeptical of a blanket ban, characterizing market regulation as "generally... a federal and provincial matter, not a city matter," and suggesting the City's real, defensible room to act is narrower: conditions on business licences (disclosure requirements), conditions on City contracts and City-run grocery initiatives, and possibly requiring licensed retailers to disclose algorithmic pricing use — not a City-wide prohibition reaching every grocer, let alone out-of-jurisdiction online platforms. A third on-record voice — from inside Council itself — concedes the point while defending the motion: Councillor Dianne Saxe (Ward 11 University–Rosedale, a lawyer and former Environmental Commissioner of Ontario) told TorontoToday: "It's a stretch for the city in terms of our legal authority, but I think that's okay. It's worth exploring what we can do because the abuse is real" (verified against the published article, July 30, 2026). Toronto's own record with COTA-derived licensing power is genuinely mixed: the City's short-term-rental bylaw (Chapter 547) was upheld by the Local Planning Appeal Tribunal in November 2019, while its ride-hailing licence-cap bylaw drew an internal City Solicitor warning that it was likely to be quashed by a court for the way it was adopted. The pattern across both precedents is that narrow, well-drafted, licensing-conditioned measures survive; broad or procedurally rushed ones do not — which maps directly onto the council item's own disclosure-to-prohibition ladder, and suggests the disclosure end of that ladder is legally safer ground for the City than the prohibition end.

Can any of this actually be enforced against platforms and data brokers outside Toronto's or Ontario's reach? Probably only partially, and this problem gets worse, not better, the higher up the ladder Toronto's ask goes. Manitoba's Bill 49 — Canada's first surveillance-pricing law, now in force since July 1, 2026 (a status verified against Manitoba's own proclamation record this review; see Chapter 6.2) — faces the identical problem one level up: counsel at MLT Aikins flagged that "enforcement against out-of-province or international e-commerce platforms may raise practical and jurisdictional challenges." The doctrine, however, is less hopeless than the practical record: the Supreme Court's "real and substantial connection" line — originating in Libman v. The Queen, [1985] 2 S.C.R. 178 and applied to provincial regulators reaching out-of-province defendants in Sharp v. Autorité des marchés financiers, 2023 SCC 29 — and the BC Court of Appeal's February 2026 ruling that Clearview AI, a foreign company with no BC presence, remained subject to BC privacy law because it collected BC residents' data (Clearview AI Inc. v. British Columbia (Information and Privacy Commissioner), 2026 BCCA 67) together establish that Canadian provincial law can, in principle, reach a foreign platform that reaches into the province. All three authorities were verified against the primary judgments this review. What no case yet establishes is the same reach for a municipality. Canadian consumer-protection enforcement generally is fragmented by province, with limited reach against out-of-province or foreign-domiciled vendors; the Federal Court's 2017 A.T. v. Globe24h.com decision shows that extraterritorial orders are legally possible under federal privacy law, but that avenue runs through the federal Privacy Commissioner and Federal Court, not through a municipality or even a province. In the United States, Instacart has directly sued New York City over a comparable municipal regulatory push, arguing federal preemption and unconstitutional interference with interstate commerce — a live illustration, albeit under different constitutional doctrine, of exactly the platform-versus-city jurisdictional fight Toronto would likely face if it tried to reach delivery platforms and data brokers directly.

What are other governments already doing? The field has moved fast in 2026. Manitoba's Bill 49 received Royal Assent June 1, 2026 (S.M. 2026, c. 41) and was proclaimed in force effective July 1, 2026 — Canada now has one binding, operating surveillance-pricing law, a fact that materially strengthens the precedent Toronto's own origin letter cites. (The master briefing's earlier "assented but not yet proclaimed" reading was correct only for the first half of June and is corrected throughout this draft; the assent date itself was also previously mis-recorded as May 5.) Ontario's opposition-sponsored Bill 104 (Fair Grocery Prices Act) passed first reading in April 2026 and is stalled awaiting second reading, and Premier Doug Ford has publicly rejected a provincial ban ("nothing beats a free market"). Federally, an NDP unanimous-consent motion to ban surveillance pricing was defeated in the House of Commons in April 2026; the government's own response runs through Bill C-36 (the "Protecting Privacy and Consumer Data Act," a PIPEDA replacement introduced June 15, 2026, currently awaiting second reading) and the June 2026 National AI Strategy and National Food Security Strategy, both of which name surveillance pricing as a harm to be addressed through strengthened privacy law rather than a direct ban. In the United States, Maryland and Connecticut (both effective October 1, 2026 — Connecticut's date corrected this review against its enrolled act; secondary reports of July 1, 2027 were wrong), and New Jersey (ban effective August 1, 2027, with its electronic-shelf-label moratorium arriving earlier, February 1, 2027) have all enacted grocery- or retail-sector surveillance-pricing restrictions; New York already has a disclosure law in force (effective July 8, 2025) and has passed (but not yet delivered to the Governor, so no signature clock is even running) a fuller ban; Colorado's governor vetoed a similar bill in June 2026; California's is pending in the state Senate. The EU relies on GDPR Article 22's automated-decision-making protections, of contested applicability to ordinary price personalization, while a dedicated Digital Fairness Act addressing "unfair personalisation and profiling, including personalised pricing" is expected to be proposed in Q3 2026. No country has yet enacted a comprehensive, sector-general ban; every enacted law found in this research is either disclosure-based, asymmetric (banning only price increases, not decreases), or sector-specific to food/grocery retail.

Who is actually harmed, and by what? The clearest, best-documented equity concern is not (yet) a proven Canadian grocery case but a structural one: the same data infrastructure that could be used to raise prices on time-poor, low-income, senior, or digitally-excluded shoppers is already documented working exactly that way in adjacent sectors — Kroger's income-predictor discount targeting in the U.S., the Meta/DOJ 2022 settlement over discriminatory ad-delivery algorithms, and well-replicated findings that algorithmic auto-insurance pricing charges higher premiums in predominantly Black neighbourhoods for comparable risk ("bluelining," per Consumer Federation of America and Greenlining Institute research, building on ProPublica's original investigation). Disclosure-only remedies are the weakest protection for exactly the consumers the motion says it wants to protect: reading and acting on a disclosure notice presumes English fluency, digital literacy, a smartphone or data plan, and time to comparison-shop — resources unevenly distributed and scarcest among the 24% of the Canadian population (9.8 million people, per University of Toronto's PROOF research initiative) living in food-insecure households in 2025.

What do critics say? The strongest counter-case, laid out fully in Chapter 8, comes from the Information Technology and Innovation Foundation (ITIF), which argues bans do not lower prices but redistribute them — pushing the low end of the price distribution up when firms can no longer offer below-average prices to price-sensitive shoppers — and cites a field experiment finding over 60% of consumers paid less under personalized pricing. ITIF and others also warn that broadly drafted bans risk sweeping in ordinary loyalty programs, senior/student discounts, and perishable-markdown systems (one cited study found automatic markdown personalization cut food waste 9.6% in treated categories), and that existing competition, human-rights, and consumer-protection law already reaches the genuine harms (collusion, discrimination, deception) without a new categorical ban.

Bottom line. Toronto's council item is a fact-finding directive, not a ban, arriving in a country where the practice it targets is capability-documented but not yet Canada-confirmed, where the City's own legal authority to act unilaterally is genuinely uncertain beyond narrow licensing-based measures, where cross-border enforcement is a real and largely unsolved problem even for enacted provincial and U.S. state laws, and where the most defensible near-term options — disclosure requirements tied to business licensing, conditions on City contracts and City-run grocery initiatives — sit at the bottom of the very ladder the motion itself proposes climbing. The comparative record assembled below gives Toronto's staff real, working statutory language to draw on at every rung of that ladder, and this report organizes it explicitly for that purpose.

Chapter 1 — The Problem and Its Documented Scale

1.1 What "surveillance pricing" means, precisely

There is no single settled legal definition. The FTC's January 2025 "Issue Spotlight: The Rise of Surveillance Pricing" deliberately surveys how the term is used across industry, press, and academia rather than adopting one, describing the practice as using "advanced data collection technologies... along with... AI and other technologies" to set individualized prices from personal data such as precise location, browsing and shopping history, and demographics (FTC, "Issue Spotlight: The Rise of Surveillance Pricing," Jan. 17, 2025; FTC press release, Jan. 17, 2025).

Canada's own Competition Bureau supplies the most citable government-source definition for this report's purposes: its June 2025 discussion paper explicitly equates "personalized pricing, also known as surveillance pricing" with pricing keyed to individual consumer data, distinguishing it from ordinary "dynamic pricing" keyed to market conditions (demand, competitor prices, inventory, time, weather) that applies the same price to everyone at a given moment (Competition Bureau of Canada, "Algorithmic pricing and competition: Discussion paper," June 10, 2025). The same paper supplies useful vocabulary this report uses throughout: first-degree price discrimination (pricing to each individual's exact willingness to pay — what "surveillance pricing" chiefly targets), second-degree (quantity/terms-based, e.g., bulk discounts), and third-degree (group-based, e.g., senior or student pricing — already normalized and explicitly not what Toronto's motion targets).

Manitoba's Bill 49 — the only enacted Canadian statute on point — supplies the most operational legal definition to date, reproduced in full in Chapter 4.

1.2 Is surveillance pricing actually happening in Toronto and Canadian grocery retail today?

Documented, in the United States:

Not documented, in Canada, as of this research:

Conclusion for this chapter: the infrastructure and commercial incentive for surveillance pricing are real and rapidly scaling in Canadian grocery (large-scale electronic shelf label rollouts, loyalty-program data assets, an active third-party pricing-software market), and the practice is documented in adjacent U.S. markets and in the closely related U.S. retail-delivery sector serving some of the same platforms operating in Canada (Instacart). But as of July 31, 2026, no Canadian regulator, journalist, or academic study has documented a Canadian grocer charging two shoppers different individualized prices for an identical item based on personal data. Toronto's motion is precautionary, not remedial of a proven local harm — the honest framing the eventual 2027 staff report should adopt, and the framing this report adopts throughout.

1.4 A platform-by-platform survey — every grocery-delivery app operating in Toronto, US-owned and Canadian-owned alike

Because the practice depends on the specific app or retailer a shopper uses, this report checked every major grocery-delivery platform known to operate in Toronto — both foreign (mostly U.S.-headquartered) and Canadian-owned — individually, rather than relying on Instacart alone. The finding holds across the board: no platform operating in Toronto, regardless of ownership, has a documented Canada-specific finding of individualized, personal-data-driven pricing.

U.S.-headquartered platforms operating grocery delivery in Toronto:

Canadian-owned platforms operating grocery delivery in Toronto:

The asymmetry finding. Multiple independent Canadian sources converge on a single, quotable conclusion. Jake Okechukwu Effoduh (assistant law professor, Toronto Metropolitan University) wrote directly: "It's difficult to know how much surveillance pricing is happening in Canada, if at all. So far, there has been no confirmed Canadian case, and the practice is opaque by design" (The Conversation / TMU News, May 4, 2026). Every specific, named, individual-price-variation finding surfaced anywhere in this research — Instacart's up-to-23% basket variation, Kroger's income-predictor score, U.S. reporting on Target's app-based location pricing — traces back to a U.S.-headquartered company, and the Consumers Council of Canada's own May 18, 2026 explainer on personalized pricing likewise cites only American examples as its evidentiary base. No Canadian-owned grocery or delivery platform surveyed here — Loblaw/PC Optimum, Sobeys/Voilà, Metro, Save-On-Foods/Pattison, or Longo's/Grocery Gateway — has a documented individual-level algorithmic-pricing finding; the closest thing to a regulatory "hit" on a Canadian-owned company is the PC Optimum data-retention finding above, which concerns data hygiene, not price discrimination. This asymmetry is itself worth stating plainly in the 2027 staff report: the documented evidence base for this entire policy debate, in Canada as in Toronto specifically, currently rests entirely on foreign (mostly U.S.) precedent, not on a confirmed domestic instance at any platform, foreign-owned or Canadian-owned alike.

1.3 The economic backdrop ("why now")

Chapter 2 — How We Got Here: The Technology and Market Forces

The FTC's Issue Spotlight identifies three converging technological shifts that make surveillance pricing possible at scale, a framework this report adopts because no better public taxonomy was found in this research:

  1. Data. A large data-broker and online-advertising ecosystem (the FTC's Spotlight situates this within a US online-advertising industry exceeding $225 billion/year) now routinely collects, aggregates, and resells behavioural, locational, and inferred-demographic data — the raw material surveillance pricing consumes (FTC Issue Spotlight, Jan. 2025).
  2. Algorithms. AI/ML systems can now predict individual willingness-to-pay from large, unstructured behavioural datasets, moving beyond the simple, rules-based, human-set-trigger pricing engines of a decade ago toward self-adjusting, harder-to-audit "black box" systems — a distinction the Competition Bureau's discussion paper draws explicitly (Competition Bureau, June 2025).
  3. Infrastructure. E-commerce personalization (product recommendation and pricing engines) and physical-store technology — chiefly electronic shelf labels (ESLs) capable of remote, centralized, near-instant price updates — now make per-customer or per-moment price variation technically trivial to deploy, whatever a given retailer chooses to do with that capability.

Electronic shelf labels in Canadian grocery are a real, large-scale, fast-moving rollout. Loblaw's ESL deployment has been described by vendor SOLUM as the largest such deployment in North American retail history (SOLUM press release). This is precisely why Toronto's Recommendation 2(d) singles out "greater transparency about electronic shelf labelling systems" as its own distinct ask, separate from the personalized-pricing question — the technology enabling both ordinary centralized price management and, potentially, individualized pricing is the same hardware, and the public cannot currently distinguish the two from the shelf.

Grocery loyalty programs are the other half of the infrastructure story. Kroger's "precision marketing" data-sales division — built on loyalty-card data covering the overwhelming majority of its transactions — reportedly generates more than 35% of the company's net income under what the company calls an "alternative profit" model, illustrating the direct commercial incentive driving investment in this infrastructure even before any pricing use is confirmed (ConsumerAffairs, May 2025). Canadian grocers operate comparable loyalty programs (PC Optimum, Scene+, Air Miles) whose data architecture is structurally identical, even though — per Chapter 1 — no Canadian grocer has been documented using this data for individualized pricing specifically.

Market concentration compounds the risk. The Competition Bureau's characterization of Canadian grocery as an oligopoly matters for algorithmic-pricing analysis specifically: the Bureau's own discussion paper notes markets with fewer firms, high price transparency, frequent repeat interactions, and high barriers to entry are the markets most vulnerable to algorithmic coordination — a "hub-and-spoke" or "rimless hub-and-spoke" pattern in which competing retailers using the same third-party pricing-software vendor can produce collusive-like pricing outcomes without any direct communication between them (Competition Bureau, June 2025, cited above). Canadian grocery — three firms controlling the large majority of sales — fits that description closely. A scale figure verified against the Bureau's own discussion paper text: "In Canada, over 60 companies offer services that use algorithms and that claim to help companies optimize pricing" — the third-party vendor layer (the "hub" in hub-and-spoke) is not hypothetical but an existing domestic industry (caveat, per the first-pass verification lane: the figure traces to a software-directory count rather than original Bureau research — the footnote did not survive outside-review extraction; browser-recheck before filing). Vendor-side evidence, verified against the companies' own announcements: Instacart acquired Eversight — the very pricing-experimentation tool at the centre of the U.S. findings in Chapter 1.2 — in September 2022 (Instacart/PR Newswire); SOLUM, vendor of Loblaw's "largest ESL deployment in North America," separately markets a general "Dynamic Pricing" retail module in its own sales collateral ("Peak Hour Pricing: Increase prices on high-demand items during lunch or weekend rush... Localized Pricing: Adjust prices by region, store, or even shelf zone—based on customer demographics") — note precisely: SOLUM does not say Loblaw uses that module, and the two claims live on different SOLUM pages; and Pricer AB's announced Sobeys deployment targeted 5 million labels across Canada by April 30, 2026, extended by a further US$51M deal signed April 20, 2026 (Pricer press releases). Using vendors' own marketing against retailers' denials is a legitimate evidentiary move for staff — provided the Loblaw-specific gap just noted is preserved, because it is exactly the kind of inferential leap a retailer's counsel would pounce on.

This is the question the council item itself treats as open (Recommendation 2 asks staff for "a jurisdictional scan and analysis of existing regulatory authorities" before proposing anything), and it deserves the same treatment here.

3.1 What the City actually voted on

The item began as a single-recommendation letter from Mayor Chow directing the City Manager to identify mechanisms and report back in Q1 2027. At Executive Committee on July 21, 2026, Councillor Bravo's motion added seven further recommendations (the jurisdictional-scan ladder, the Ontario ask, the federal-collaboration ask, and the ISED/Competition Bureau/Privacy Commissioner advice requests, plus the AMO/FCM forward), which carried, and Chow's motion to adopt as amended also carried. That amended, eight-recommendation item is what went to and passed Council July 29–31, 2026 (City of Toronto, Agenda Item 2026.EX33.33; origin letter, backgroundfile-289756.pdf). Committee-level votes are recorded only as "Carried," with no numeric division. The full Council vote is now posted on the item page: Adopt Item — Carried by a recorded vote of 26–0 (0 absent), July 30, 2026, adopted without further amendment, with the City's standard caution that Council decisions remain preliminary until the meeting minutes are confirmed (Agenda Item 2026.EX33.33, retrieved August 5, 2026). Earlier versions of this report (v1.0–v1.1) noted the recorded-vote form could not yet be independently confirmed; the posted record has since resolved the question in favour of a recorded, unanimous vote.

Verified this review — the July motion was not Council's first word on this subject. On March 25–26, 2026, Council adopted (as amended) item MM39.27, "Grocery Store Pilot Project" (Perruzza/Colle), whose Decision 2(a) — added by Mayor Chow's amending motion — already directs a Q2 2027 report on "policy levers available to the City to prevent price gouging by grocery and other retailers, including ensuring retailers licensed by the City are transparent about rates and prices and disclose the use of consumers' personal data and algorithmic pricing," with 2(b) addressing grocers' anti-competitive property covenants (2026.MM39.27). Two consequences: the disclosure-via-licensing rung of this report's ladder already has a live Council directive behind it predating 2026.EX33.33; and staff now owe two overlapping reports (Q1 and Q2 2027) that should be reconciled rather than drafted in parallel ignorance of each other.

The origin letter states, verbatim: "The City of Toronto Act provides authority to regulate for the purpose of consumer protection and the economic well-being of Torontonians. This motion directs staff to explore every legal mechanism within the City of Toronto's power to ban surveillance pricing on groceries" (backgroundfile-289756.pdf). The claim is made in general terms — no specific section of the City of Toronto Act, 2006 is cited in the origin letter or the agenda item.

This review closes that gap. The relevant provisions of the City of Toronto Act, 2006, S.O. 2006, c. 11, Sched. A, verified verbatim against the current Ontario e-Laws consolidation (ontario.ca/laws/statute/06c11) on August 1, 2026 (the section numbers were first flagged by one of this project's multi-AI capture submissions and then confirmed against the Act itself — see Appendix I):

What this changes, and what it does not. The skeptical academics quoted in 3.3 below frame the problem as market regulation being "a federal and provincial matter, not a city matter" — a constitutional-culture observation that remains true as far as it goes. But the statutory text above means the City's position is stronger than "no cited authority": consumer protection is an express sphere of Toronto by-law jurisdiction, with prohibition mechanics attached. The genuinely contestable questions are (i) whether a surveillance-pricing by-law would survive a s. 11 conflict/frustration attack given the provincial and federal schemes already occupying adjacent ground, and (ii) the territorial-reach problem of 3.6. The City's most practically defensible tool remains its licensing power (used, for example, to require price transparency in the towing industry, cap payday-loan storefronts per ward, and bar pet stores from puppy-mill sourcing) — a business-licensing authority that reaches Toronto-licensed businesses, not a general power to regulate prices or data practices as such.

3.3 The skeptical case

Three legal academics quoted in a single Canadian Press article (syndicated via CHCH, QP Briefing, Insauga, and the Canadian Press newswire, July 22–23, 2026) supply the most direct, on-the-record skepticism found in this research:

(Source for the Salazar and Cohen quotes: "Toronto may be out of its depth on proposed surveillance pricing ban, experts say," The Canadian Press, July 22, 2026, via CHCH News. Sources for Effoduh and Saxe: the TorontoToday.ca article above, fetched and read in full via live browser rendering on August 1, 2026 — Saxe's quote and Effoduh's "modest things" assessment are now verbatim-confirmed; Effoduh's longer reconstructed quotes retain the earlier caution.)

The constitutional backdrop, now properly cited. The controlling modern statement of Toronto's subordinate status is Toronto (City) v. Ontario (Attorney General), 2021 SCC 34 — the Bill 5 ward-cutting case, where a 5–4 Supreme Court upheld the Province's mid-election restructuring of Toronto's council and confirmed that provinces may reshape municipal institutions "at any time," with unwritten constitutional principles unavailable to shield a city. (One capture-wave submission cited this proposition to "2021 ONSC 6001"; verification found that citation belongs to an unrelated employment case, and the correct authority is 2021 SCC 34 — see Appendix I.) The doctrine cuts precisely: it establishes that Queen's Park could override or extinguish any Toronto surveillance-pricing by-law at will — Salazar's point — but it does not say the City lacks authority to act until overridden. Toronto's by-law record on both sides of that line is set out in 3.4.

A relevant, currently favourable fact for the City: Ontario's own government has, as of this writing, declined to regulate in this space — Premier Doug Ford rejected both the Ontario NDP's non-binding motion and the Liberal-sponsored Bill 104, saying "there's no better way of letting people get lower costs... than competition," and that "nothing beats a free market," while promising to "tear to shreds" any proven collusion (CP24, "Doug Ford nixes idea of grocery surveillance pricing ban in Ontario," April 16, 2026; The Globe and Mail [deep URL not banked — flagged per source rules]). This reduces (for now) the immediate risk of a direct provincial-paramountcy conflict, since the province is not actively occupying the field — but it is a fragile basis for municipal authority, since it could change the moment Queen's Park changes its mind, and Toronto's own Recommendation 3 is explicitly asking it to.

3.4 What Toronto's own precedent suggests

Two past tests of Toronto's COTA-derived licensing power point in different directions and, read together, suggest a workable rule of thumb:

The pattern: narrowly scoped, carefully drafted, licensing-conditioned rules survive; broadly scoped or procedurally rushed rules invite successful challenge. Applied to surveillance pricing, this suggests Toronto's legally safest path runs through business-licensing conditions (disclosure requirements on licensed grocers, conditions on City contracts and City-run grocery/public-market initiatives) rather than a blanket city-wide prohibition — which is, not coincidentally, close to where the council item's own Recommendation 2(a) (disclosure) sits relative to 2(b) (full prohibition) on its own ladder.

3.5 This report's assessment

Taking both sides together: on the strength of the licensing precedents, the City can require disclosure of algorithmic pricing as a condition of a business licence, bar the practice in City-run grocery initiatives and City contracts, and lobby upper levels of government (which is what Recommendations 3–8 actually do). The statutory foundation is stronger than the July commentary suggested — consumer protection is an express s. 8(2) sphere with prohibition mechanics under s. 8(3), so "the Act doesn't say consumer protection" is no longer an available objection. But the City's authority to impose a comprehensive, city-wide prohibition reaching every grocer — and especially reaching online retailers and delivery platforms with no fixed Toronto premises — remains genuinely uncertain on the two grounds that survive the statutory find: s. 11 conflict/frustration with provincial and federal schemes, and territorial reach (3.6). It remains contested by named legal academics and conceded as "a stretch" by a councillor-lawyer who voted for it. This is not a reason for staff to abandon the ladder; it is a reason the 2027 report should be explicit that the disclosure and City-contract-conditions rungs rest on the strongest available municipal footing today (a footing the City Solicitor should still be asked to confirm), while the full-prohibition rung likely requires provincial or federal legislation to be durable against challenge — precisely the ask Recommendations 3–7 already make of Ontario and Ottawa.

One further datum for the "what can a city actually do" question, verified this review: New York City's Local Law 144 of 2021 — bias audits and candidate notices for automated employment decision tools, enforced by a City agency (DCWP) since July 5, 2023 — is a working example of a major North American city regulating algorithm use through local law. Employment, not pricing; but a direct existence proof for municipal algorithmic-accountability regulation (NYC DCWP).

3.5A The s. 11 conflict question, analyzed against the governing doctrine

Legal analysis for policy purposes, not legal advice; the City Solicitor's opinion controls for Council. Every authority below was verified against the primary judgment or consolidated statute on August 1, 2026.

The doctrine. COTA s. 11 (and its Municipal Act, 2001 twin, s. 14) codifies the Supreme Court's two-question conflicts test: (1) is dual compliance impossible; (2) does the by-law frustrate the purpose of the provincial or federal Act. The Ontario Court of Appeal applied exactly this framework to a Toronto by-law in Croplife Canada v. Toronto (City) (2005), 75 O.R. (3d) 357, at para. 74: "The appellant concedes that it is not impossible to comply with the pesticide by-law at the same time as the federal PCPA or the Ontario Pesticides Act... the by-law is not rendered inoperative by the conflicts test in s. 14 of the Municipal Act, 2001, applied in accordance with the Supreme Court's decisions in Spraytech and Rothmans" — and upheld the by-law. The governing formulations: 114957 Canada Ltée (Spraytech) v. Hudson (Town), 2001 SCC 40, at para. 38 — "A true and outright conflict can only be said to arise when one enactment compels what the other forbids" — and para. 37: a by-law is not ineffective merely because it "enhances" the statutory scheme "by imposing higher standards"; Rothmans, Benson & Hedges v. Saskatchewan, 2005 SCC 13, at para. 23: a stricter law that "simply prohibits what Parliament has opted not to prohibit" neither conflicts with nor frustrates the federal scheme; and Canadian Western Bank v. Alberta, 2007 SCC 22, at paras. 74–75: there is no occupied-field presumption — "[a]n incompatible federal legislative intent must be established by the party relying on it," under a standard of judicial restraint.

The application. A Toronto grocer (or platform) subject to a municipal disclosure duty or even a municipal prohibition can comply with it and with the Consumer Protection Act, 2002 and Competition Act simultaneously — obeying the stricter rule violates nothing in either statute, so no Spraytech para. 38 conflict arises. On frustration, the statutory record — checked verbatim this review — is silence: the consolidated CPA 2002 contains no provision on algorithmic or personalized pricing (its nearest provisions, s. 14(2) para. 11 on false price-advantage claims and s. 15(2)(b) on unconscionable gross overpricing, complement rather than cover a pricing rule); the CPA 2023 is not in force and equally silent; the Competition Act contains no personalization provision at all, and Parliament repealed its criminal price-discrimination offence (s. 50) in 2009 — a deliberate withdrawal that makes it very hard for a challenger to establish, against Canadian Western Bank's restraint standard, that a stricter municipal rule frustrates a purpose neither legislature has expressed.

The honest limits. Three, and they matter: (i) this analysis holds only while Queen's Park stays silent — if Ontario legislates (Bill 104 or otherwise), the conflict analysis reopens, and the Province can in any event simply override the City (Toronto (City) v. Ontario (A.G.), 2021 SCC 34); (ii) s. 11 doctrine protects a by-law's validity, not its reach — nothing here touches the territorial enforcement problem of 3.6; and (iii) the stronger attack on a full prohibition is likely not s. 11 at all but Eng-style absence of municipal purpose (3.4), precisely because Chapter 1 finds no documented Toronto instance of the practice — which is why the disclosure and licensing rungs, grounded in the transparency interests of Toronto consumers today, are the defensible end of the ladder — though the Eng exposure is a matter of degree, not kind: a disclosure by-law rests on the same not-yet-local-harm premise, just with a far more modest claim on municipal purpose — while the prohibition rung remains exposed on municipal purpose and reach even though it survives the conflict analysis better than the July commentary implied.

3.6 The enforcement problem, separately from the authority question

Even a legally sound municipal or provincial rule faces a second, distinct problem: reaching businesses that operate wholly or partly outside Toronto's or Ontario's territory.

Practical implication for Toronto: any rule that reaches only Toronto-licensed, Toronto-premised grocers will miss a large and growing share of the grocery-delivery and online market (Instacart, DoorDash-style delivery of grocery orders, and any data broker supplying pricing-relevant data from outside Ontario). Meaningful enforcement against that share of the market realistically requires provincial or, more robustly, federal action — reinforcing why Recommendations 3 through 7 route the harder enforcement questions upward rather than attempting to solve them at the municipal level.

Chapter 4 — The Options Ladder, With Real Model Legislative Language

This chapter builds the working ladder Toronto's Recommendation 2 asks for, from investigation-only through full prohibition. Every rung below is backed by real, named, dated law or regulation — enacted, pending, or (in one case) vetoed, each labeled precisely. No language here is invented; where exact statutory text could not be independently verified in this research pass, that is stated rather than papered over.

Rung 0 — Investigation and information-gathering only (no binding rule)

Models: U.S. FTC Section 6(b) study (2024–25); Competition Bureau of Canada discussion paper and consultation (2025–26).

The FTC's 6(b) authority let it compel eight companies (Mastercard, Revionics, Bloomreach, JPMorgan Chase, Task Software, PROS, Accenture, McKinsey & Co.) to produce information on their surveillance-pricing products, without any accompanying rule or enforcement action — a pure fact-finding exercise (FTC, July 24, 2024). Canada's Competition Bureau discussion paper and subsequent "What We Heard" report (103 responses; Jan. 22, 2026) is the direct domestic analogue, explicitly disclaiming that it reflects the Bureau's own legal position: "This report summarizes what we heard from stakeholders, and it does not necessarily reflect the Bureau's views" (Competition Bureau, Jan. 22, 2026). This is, functionally, the rung Toronto's own Recommendation 1 sits on today — a directive to identify mechanisms and report back, not yet a rule.

Rung 1 — Mandatory plain-language disclosure only

Model (enacted, in force): New York General Business Law § 349-a, the Algorithmic Pricing Disclosure Act — effective July 8, 2025, per the Attorney General's own filing in NRF v. James, with enforcement practically underway since the October 8, 2025 first-instance dismissal (an 'in force November 10, 2025' date carried from secondary summaries was corrected on outside review).

Requires an on-screen statement whenever algorithmic/personal-data pricing is used: "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA." Defines "personal data" broadly as "any data that identifies or could reasonably be linked, directly or indirectly, with a specific consumer or device." Civil penalties up to $1,000 per violation. Litigation status, verified this review: the retail industry's First Amendment compelled-speech challenge (National Retail Federation v. James, S.D.N.Y.) was dismissed by Judge Jed Rakoff on October 8, 2025; NRF's appeal is pending at the Second Circuit (briefed January–February 2026, no appellate decision as of August 1, 2026) — so "in force and upheld at first instance, on appeal" is the precise formulation, not "survived challenge" full stop (Skadden, Jan. 20, 2026; WLF case page). (Caution for staff: do not confuse §349-a with N.Y. Gen. Bus. Law §340-b, an algorithmic rent-coordination ban whose enforcement is stayed pending RealPage v. James — one capture-wave submission misrepresented §340-b as an in-force grocery full ban, quoting statutory language that does not exist; see Appendix I.)

Complementary model: Manitoba's Bill 49 deems use of "personalized algorithmic pricing" a material fact that must be disclosed to the consumer (new s.2(4) of the Business Practices Act) — a disclosure obligation nested inside a broader statute rather than a freestanding one (full text under Rung 4 below).

This is the single most directly transplantable rung for Toronto, since it maps exactly onto Recommendation 2(a) ("mandatory plain-language disclosure"), sits at the bottom of the legal-risk spectrum discussed in Chapter 3, and can plausibly be implemented via a business-licensing condition rather than requiring provincial legislation.

Model (enacted and in force since July 1, 2026): Manitoba Bill 49 (S.M. 2026, c. 41), s.2(5) — the "material fact" disclosure of Rung 1 is paired with a genuine two-part consent test, not mere fine-print notice:

For the purpose of clause (3)(q) as it relates to a supplier that is an online retailer or online distributor, a consumer is deemed to give their express consent to the higher price only if (a) the supplier prominently discloses to the consumer the reason for the higher price in clear and understandable language; and (b) the consumer takes clear overt action to consent to the higher price.

(True statutory text, s. 2(5), substituted on outside review for an earlier quote-formatted paraphrase.) Scope disclosure that matters for transplantation: as enacted, this consent standard applies to online retailers and distributors — a Toronto or Ontario version reaching in-store electronic-shelf-label contexts would need to broaden it expressly.

This is meaningfully stronger than a passive on-screen label: it requires an affirmative, comprehensible, and specific disclosure at the point the higher price is charged, plus an active consumer action — closer to an opt-in regime than a notice regime. (Full bill text: Manitoba Legislative Assembly, web2.gov.mb.ca/bills/43-3/b049e.php.)

Rung 3 — General automated-decision-making transparency (adjacent, not pricing-specific)

Models: EU GDPR Article 22; Australia's Privacy and Other Legislation Amendment Act 2024 (new Australian Privacy Principles 1.7–1.9, in force from December 10, 2026).

GDPR Article 22 grants a right not to be subject to a decision "based solely on automated processing... which produces legal effects... or similarly significantly affects" the individual, with mandatory safeguards (human review, right to contest) where an exception applies (Article 22 GDPR, gdpr-info.eu). Legal scholarship is split on whether ordinary price personalization clears Article 22's "similarly significant effect" threshold — Frederik Zuiderveen Borgesius and Joost Poort's 2020 analysis in Information & Communications Technology Law takes "a sceptical view" that most personalized pricing qualifies (Borgesius & Poort, 2020). Australia's new APP 1.7–1.9 will require that any automated decision-making system be "explainable and empirically justified," a general transparency rule not specific to pricing — wording per a secondary summary; the primary amendment text was not independently quoted in this research (Lexology summary, 2026).

A non-Western model at this rung — verified this review, filling a gap the master briefing itself flagged: China's Personal Information Protection Law, in force November 1, 2021, Article 24 (Stanford DigiChina translation): "When personal information handlers use personal information to conduct automated decision-making, the transparency of the decision-making and the fairness and justice of the handling result shall be guaranteed, and they may not engage in unreasonable differential treatment of individuals in trading conditions such as trade price," paired with a mandatory non-personalized option or convenient refusal mechanism for algorithmic marketing (DigiChina, PIPL translation). This is the closest thing anywhere to a statutory prohibition on algorithmic price discrimination in a major economy — the anti-"big data ripening" (大数据杀熟) provision — enacted five years before any North American equivalent. Honest caveats: the translation is unofficial, and no publicly documented enforcement action specifically under Art. 24 against price discrimination was located as of August 2026 — a prohibition on paper whose enforcement runs mainly through adjacent platform-regulation instruments.

Assessment: this rung is useful as a backstop (a general transparency default that would incidentally cover pricing algorithms) but is not a purpose-built solution and its applicability to routine price personalization is legally contested even in the EU, where it has existed since 2018. The PIPL comparison cuts the other way: a purpose-built prohibition can be enacted and still lie dormant without an enforcement apparatus behind it — the design lesson for Toronto being that the enforcement machinery question (Chapter 5) is not separable from the rung choice.

Rung 4 — Asymmetric, sector-specific prohibition (bans price increases only, via personal data)

Model (enacted and in force): The Business Practices Amendment Act, S.M. 2026, c. 41 (Bill 49) — assented to June 1, 2026; in force July 1, 2026 by proclamation (proclamation signed June 17, 2026 — verified this review against the King's Printer of Manitoba's proclamation record and the consolidated C.C.S.M. c. B120, which already incorporates the amendments) — amending The Business Practices Act, C.C.S.M. c. B120. This is now a binding, operating law, not a pending one: the only in-force surveillance-pricing statute in Canada, and the single most-litigated fact of this project's entire multi-AI capture wave (see Appendix I — the master briefing's "not yet proclaimed" reading was correct when written and was overtaken by events; several capture tools called it "in force" for the wrong reasons before it was true).

Exact statutory definition of "personalized algorithmic pricing" (added to s. 1 by the amending Act — verbatim, fetched from the Legislative Assembly of Manitoba's own bill page this review; an earlier normalized paraphrase mislabeled "verbatim" was corrected on outside review):

"personalized algorithmic pricing" means, subject to the regulations, pricing that is based on the use of an algorithm or automated processing to set, recommend or vary a price offered to a specific consumer as a result of data about the consumer collected, analyzed or processed with or without the consumer's consent, knowledge or involvement, including, without limitation, data about (a) the consumer's personal information, attributes and behaviours, such as (i) the consumer's browsing or purchasing history, consumer habits or spending patterns, (ii) the consumer's electronic devices used in browsing or purchasing and their profiles on such devices, (iii) the consumer's inferred willingness to enter into the consumer transaction, (iv) the consumer's demographics or socio-economic status, including their income level, (v) the consumer's employment pay period or financial assistance payment schedule, (vi) the consumer's credit history, (vii) the consumer's location, including their address for the delivery of the good, (viii) the consumer's medical history or health status, and (ix) any other information, attribute or behaviour prescribed by regulation, and (b) any other matter prescribed by regulation;

Note the two limbs the paraphrase had dropped: the whole definition is "subject to the regulations," and clause (b) is an open regulation-making limb beyond the (a)(ix) data-category power — Manitoba built itself room to extend the definition without returning to the legislature.

New unfair-business-practice clauses (s.2(3)): (r.1) demanding a higher price at point of sale via an electronic shelf label due to personalized algorithmic pricing; (v) for online retailers/distributors, using personalized algorithmic pricing to increase the price charged. Note precisely what this does not do: it does not ban dynamic/market-condition pricing generally, does not touch loyalty/rewards discounting, and does not reach price decreases — only data-driven price increases. Penalties on summary conviction, per Blakes' analysis of the Act's existing penalty provisions: up to $300,000 (first offence) and up to $1,000,000 (subsequent offences) for corporations, plus possible consumer restitution orders — the amending Act itself touches only the limitation period (new s. 33(3)), so these figures rest on the pre-existing s. 33, not independently verified this review (Blakes, "Manitoba Proposes to Ban Personalized Pricing Under Bill 49," April 2, 2026). Full text: web2.gov.mb.ca/bills/43-3/b049e.php.

Directly transplanted model, still pending: Ontario's Bill 104 (Fair Grocery Prices Act, 2026, sponsored by Liberal MPP John Fraser, first reading April 15, 2026, ordered for second reading, not yet passed) uses a virtually identical clause-for-clause definition of "personalized algorithmic pricing," amending both the Consumer Protection Act, 2002 and Consumer Protection Act, 2023 to add: "It is an unfair practice to use personalized algorithmic pricing to inform a change in price for an individual consumer," with named examples of electronic shelf labels and online-platform price variation (Legislative Assembly of Ontario, Bill 104). This is the ready-made model for Toronto's Recommendation 3 ask to Queen's Park — it already exists, in bill form, and needs only political will (currently absent — see Chapter 3) to advance past first reading.

Rung 5 — Sector-specific full prohibition (food retail), with real, documented enforcement gaps

Model (enacted, effective October 1, 2026): Maryland HB 895, the Protection From Predatory Pricing Act, signed by Gov. Wes Moore April 28, 2026 (Chapter 154 of 2026 Acts). Bill-number conflicts resolved this review against the Maryland General Assembly record: HB 895 is the sole enacted vehicle; its cross-file was SB 387 (not "SB 597"), and it died in the House; and a capture-tool cite to "Agriculture §13-408" confused the provision with Commercial Law §13-408 (the MCPA private-action section, amended precisely to exclude these violations — see below).

New Md. Com. Law § 13-321 creates two independent prohibitions for "food retailers" (establishments ≥15,000 sq. ft. selling tax-exempt food) and "third-party delivery service providers": (1) a ban on setting a personalized price for "a specific consumer" based on personal data; (2) a ban on using personal data to charge "a single consumer" higher prices than another consumer; plus a separate ban on using data proxying for protected-class membership to withhold an advantage available to others. "Personal data" is defined by cross-reference to Md. Com. Law § 14-4701 as "any information that is linked or can be reasonably linked to an identified or identifiable consumer."

Documented loopholes, per IAPP's legal analysis — instructive for what Toronto should avoid replicating: no defined "baseline price" for comparison; the statute bans pricing "a specific consumer" or "a single consumer," arguably leaving group or segment-based personalization unreached (broader "group of consumers" language was deliberately stripped before passage); a loyalty/membership/rewards-program carve-out exempts the primary vector through which granular personal data is actually collected in the first place; and there is no private right of action — §13-321(E) says so expressly, and the enrolled bill amends Com. Law §13-408 to carve these violations out of the MCPA's private-action route; enforcement runs through the Maryland Attorney General's Consumer Protection Division, with a mandatory 45-day cure notice before any action. Two claims from the capture wave did not survive verification against the enrolled text: there is no "prices must stay fixed at least one business day" rule (that language was struck before passage), and the "$10,000/$25,000" penalty figures are the MCPA's pre-existing generic civil penalties (Com. Law §13-410), not bespoke caps written into this act (enrolled bill, hb0895e.pdf; IAPP, May 13, 2026; Skadden, May 8, 2026).

Rung 6 — General-retail hybrid: prohibition plus a disclosure fallback

Model (enacted, in force October 1, 2026 — date corrected this review against the enrolled act; secondary reporting of July 1, 2027 was wrong): Connecticut Public Act No. 26-64 (substitute Senate Bill 4 — same law), signed by Gov. Ned Lamont May 27, 2026, the third amendment to the Connecticut Data Privacy Act. The pricing provisions are § 11 of the act (confirmed against the enrolled PDF this review), effective October 1, 2026.

Structure — broader than Maryland's, and a genuinely graduated model, now quoted from the enrolled text: (1) an outright prohibition, § 11(c)(1): "no retail seller or third-party delivery service doing business in the state shall engage in surveillance pricing"; with exceptions at § 11(c)(2) for retention discounts, "justifiable differences in the costs incurred... including... consumer selections, delivery distances or delivery times," "justifiable temporal differences... based on supply and demand," and publicly disclosed uniform-terms discounts for broadly defined groups (veterans, seniors, students, teachers, residents of a specific area) and opt-in loyalty/membership/rewards programs — note there are no "pricing errors" or "outage" exceptions (that list belongs to New York's statute; earlier drafts conflated the two); § 11(d) exempts insurance licensees, GLBA financial institutions, and banks; plus (2) a mandatory disclosure requirement, § 11(b)(1), whose exact string is "THIS PRICE WAS INCREASED BY A PRICE SETTING DEVICE USING YOUR PERSONAL DATA" (or a substantially similar disclosure, "readily visible to the average consumer") — a longer string than New York's, and previously misquoted in this project's own drafts. Enforcement: § 11(e) makes violations CUTPA offences, enforced solely by the Attorney General; no private right of action (enrolled act PDF).

This two-tier prohibition-plus-disclosure-fallback structure is arguably the single best template for Toronto's own Recommendation 2, since Recommendation 2 itself asks for a ladder running "at minimum" disclosure "up to" full prohibition within one report — Connecticut has already built exactly that as one integrated statute, sector-general rather than food-only. Its added significance, post-correction: with an October 1, 2026 commencement it will be the earliest-operating general-retail surveillance-pricing regime in North America — an in-force template, not a pending one.

Rung 7 — Full prohibition with a private right of action

Model (enacted; staggered commencement — verified this review against the act's own §9): New Jersey's Fair Price Protection Act (Assembly Committee Substitute for A4085/A4523; Senate companion S3612), signed by Gov. Mikie Sherrill July 23, 2026 — the most recent, and on enforcement grounds the strongest, of the enacted U.S. state laws.

Bars using personal data (purchase history, online search history, location) to set individualized higher grocery prices. Commencement, corrected from earlier reporting: the core pricing ban (§3) takes effect August 1, 2027 (first day of the 13th month after enactment); the one-year moratorium on new ESL installations (§4) takes effect February 1, 2027 (repair/replacement of existing units exempt); the ESL impact study (§5) is effective immediately — so the widely reported "February 1, 2027" date belongs to the ESL moratorium, not the ban. Enforcement, stated precisely: violations are "unlawful practices" under New Jersey's Consumer Fraud Act, which means the CFA's existing private action and treble-damages route attaches (rather than the act creating a standalone new private right, as EPIC's shorthand suggested), and the act adds its own AG civil action for the greater of actual damages or $50,000 (§7). The ESL-moratorium-pending-study design remains directly on point for Toronto's Recommendation 2(d) (A4085 second reprint, pub.njleg.gov; Governor's release, July 23, 2026; EPIC, July 23, 2026).

Pending model, not yet law: New York's One Fair Price Act (S.8623-B/A.9349-B) passed both legislative chambers in early June 2026 but — corrected this review against the Senate's bill-status record — has not yet been delivered to Governor Hochul, and New York's constitutional signing clock (10 days Sundays excepted in session; 30 days out of session) runs from delivery, not passage, so no signature deadline currently exists; the widely repeated "December 31, 2026 deadline" is not a statutory fact. It would ban using personal data — purchase history, browsing history, real-time location, income, inferred household size — to set individualized prices, while expressly preserving legitimate discounts, coupons, loyalty pricing, and senior pricing; its enforcement is Attorney-General-only (new § 349-a subd. 4: AG action "in addition to any other lawful remedy available"; an earlier claim that it carries a private right of action was corrected on outside review against the passed B-text), with penalties up to $5,000 (first violation) / $20,000 (subsequent) (EPIC, June 4, 2026; Wilson Sonsini). Its explicit carve-outs for loyalty/coupon/senior pricing directly answer the "definitional overbreadth" objection industry critics raise (Chapter 8) and could usefully be copied into any Toronto- or Ontario-drafted prohibition.

Rejected model, instructive as a boundary case: Colorado's HB 26-1210 passed the state legislature in early May 2026 but was vetoed by Gov. Jared Polis on June 2, 2026. It would have banned personal-data-based individualized pricing and algorithmic wage-setting in one statute, with a private right of action and statutory damages — the broadest and most aggressive model found in this research, and the only one a governor has affirmatively rejected (EPIC, "Colorado Governor Vetoes Surveillance Pricing Bill," June 4, 2026). The veto letter's reasoning is now verified (June 2, 2026, as quoted by Colorado Newsline from the Governor's letter): "Instead of specifically defining and targeting unethical conduct and practices, the bill takes a broader approach to capture any technology that incidentally influences a price or wage amount... Because of the broad sweep, the bill would punish differentially lower prices, not just higher prices," and "In practice, this means that many Coloradans won't get discounts on items they buy if I were to sign this" (Colorado Newsline). The Colorado–Maryland pairing is therefore the ladder's cleanest verified design lesson: Colorado died of definitional overbreadth; Maryland survived by banning only data-driven increases and exempting discounts. Any Toronto or Ontario drafting exercise should be tested against exactly this veto letter.

Also pending, broadest scope: California AB 2564 (the "Surveillance Pricing Prohibition Act"), introduced February 20, 2026, passed the Assembly June 4, 2026; cleared the Senate Privacy Committee 5–2 on June 22, 2026 and sits in Senate Judiciary as of August 1, 2026 (status verified this review). Would ban surveillance pricing across grocery, brick-and-mortar, and e-commerce retail generally (not sector-limited), including data sourced from third parties; civil penalties up to $12,500 per violation, tripled for intentional violations (Crowell & Moring; California Legislative Information).

Rung 8 — Electronic shelf label transparency, as a standalone requirement

Models: Manitoba Bill 49's ESL-specific unfair-practice clause (Rung 4, above) and New Jersey's one-year ESL installation moratorium (Rung 7, above) are the only two enacted instruments found that address ESLs by name rather than folding them into a general pricing-algorithm definition. Toronto's Recommendation 2(d) asks for exactly this as a distinct, severable component — meaning the City could plausibly pursue ESL-specific disclosure or a temporary moratorium on new installations even if a broader personalized-pricing prohibition proves legally unreachable at the municipal level, since ESL transparency is arguably closer to a pure business-licensing/point-of-sale-labelling matter than a data-practices prohibition.

Rungs above prohibition — the upstream tier the ladder currently stops short of

An architectural framing contributed by the capture wave — its factual anchor now fully verified — the strongest structural contribution of the whole harvest is the observation that every rung above stops at the pricing conduct while leaving the data supply chain untouched — and that a genuinely complete ladder has at least two rungs higher: (i) restrictions on the collection/retention of the specific data categories that feed pricing engines, and (ii) data-broker registration and audit regimes of the kind Connecticut has begun building alongside its pricing law. On this framing, even a full prohibition (Rung 7) is a downstream remedy; the upstream rungs are the only ones that reach the data brokers Chapter 5.3 identifies as the least-reachable actors. The Connecticut anchor is confirmed from the enrolled act itself: PA 26-64 §§ 1–10 (effective October 1, 2026) build the data-broker regime alongside § 11's pricing rules — no selling or licensing personal data unregistered on or after January 1, 2027 ($2,500 fee), an accessible deletion mechanism by July 1, 2028, verification duties August 15, 2028, 45-day broker compliance checks from October 1, 2028, annual public statements by July 1, 2029, triennial independent audits from July 1, 2031, and civil penalties up to $200/day. The architectural point is no longer just the wave's framing — one U.S. state has already legislated the pricing rung and the data-supply rung as a single package.

4.1 How the ladder maps onto Toronto's own Recommendation 2

Toronto's ask (Rec. 2)Nearest real-world modelStatusLegal risk at municipal level (Ch. 3)
(a) Mandatory plain-language disclosureNY Gen. Bus. Law §349-a; Manitoba s.2(4)Enacted & in force (NY since July 8, 2025; MB since July 1, 2026)Lowest — closest fit to licensing-condition authority
— (consent standard, not separately asked but a natural bridge)Manitoba s.2(5) two-part consent testEnacted & in force (since July 1, 2026)Low-moderate
(c) Opt-out / restriction on predatory data practicesManitoba's unfair-practice designation; CT's prohibition-with-exceptions structureEnacted & in force (MB July 1, 2026; CT Oct. 1, 2026)Moderate — likely needs provincial backing to be durable
(b) Full prohibition of price-raising algorithmic pricingMaryland (food-specific); Connecticut (general, hybrid); New Jersey (food-specific + PRA); NY One Fair Price Act (pending, general)Mixed — 3 enacted, 1 pendingHighest — this report's Ch. 3 assessment is that a City-wide version is legally vulnerable without provincial or federal legislative backing
(d) Electronic shelf label transparencyManitoba ESL clause; NJ moratoriumEnacted & in force (MB); enacted, in force Feb. 2027 (NJ)Low — most severable, closest to a pure point-of-sale labelling rule

Recommendation embedded in this chapter: the legally strongest, fastest-to-implement package available to Toronto today combines Rung 1 (disclosure) and Rung 8 (ESL transparency) as business-licensing conditions, while Rungs 4 through 7 (the actual prohibition rungs) are the ones that should be directed at Queen's Park under Recommendation 3 — using Ontario's own Bill 104, already drafted and modeled on Manitoba's enacted law, as the ready-made vehicle, rather than Toronto attempting to legislate a citywide prohibition of doubtful vires.

Chapter 5 — The Levers of Power: Who Decides, Who Pays, Who Can Enforce Against Whom

5.1 Who decides, at each level

Toronto: the City Manager, City Solicitor, and Executive Director of Municipal Licensing and Standards (Recommendation 1); the ED of Municipal Licensing and Standards jointly with the General Manager of Economic Development and Culture (Recommendation 2). Council itself must ultimately adopt whatever bylaw or licensing amendment staff propose in Q1 2027 — this report finds no indication staff have pre-committed to any specific rung of the ladder.

Ontario: the Legislative Assembly, currently controlled by Premier Doug Ford's Progressive Conservatives, who have publicly rejected a ban; the only currently drafted vehicle (Bill 104) is a Liberal private member's bill without government support and has not passed second reading. The Ministry responsible for the Consumer Protection Act (Public and Business Service Delivery and Procurement) would administer any enacted provincial rule.

Federal Canada: Parliament, via Bill C-36 (first reading completed June 15, 2026; awaiting second-reading debate, no votes recorded — LEGISinfo, verified Aug. 1, 2026); the Competition Bureau (an independent law-enforcement agency, not a regulator empowered to set prices — see 5.3 below); the Office of the Privacy Commissioner of Canada (whose private-sector enforcement powers would transfer to a new body, the Digital Safety and Data Protection Commission of Canada, if and when Bill C-36 passes — confirmed directly by the Privacy Commissioner's own June 15, 2026 statement on the bill, priv.gc.ca).

5.2 Who pays

5.3 Who can enforce against whom — the jurisdictional reality check

Chapter 6 — What Other Governments Are Already Doing

The comparative record below covers every jurisdiction identified in this research with either enacted law, a pending bill, a rejected/vetoed bill, or a formal government study specifically on surveillance/algorithmic personalized pricing.

6.1 Canada — federal

InstrumentStatusKey dateWhat it does
PIPEDA (current)In force; silent on surveillance pricing—No provision addresses algorithmic/personalized pricing; general consent framework only
Bill C-36 (Protecting Privacy and Consumer Data Act)First reading completed; awaiting second-reading debate, no votes recorded (LEGISinfo, verified Aug. 1, 2026)Introduced June 15, 2026Replaces PIPEDA; creates new Digital Safety and Data Protection Commission; penalties up to $25M/5% global revenue. Definitional text now verified verbatim against the first-reading bill: s. 2(1) — "personal information means information about an identifiable individual, including information that is inferred about the individual"; s. 2(2) — "For greater certainty, de-identified personal information does not cease to be personal information." The "inferred information" clause is the statutory hook that reaches willingness-to-pay inference models even though the bill never names surveillance pricing. Tabling-day framing, verified (The Globe and Mail, June 15–16, 2026 — one tool misattributed this to iPolitics): Minister Solomon said the bill "takes aim at surveillance pricing to stop price gouging" while sparing beneficial data uses, "for example for loyalty cards," with the boundary left to regulator guidance; the CCLA's verified critique warns the bill lets businesses "ignore consent requirements and retain personal data indefinitely if superficially de-identified" (CCLA)
Bill C-226 (National Framework for Food Price Transparency Act; PMB, Gurbux Saini, Lib.)Passed second reading 168–150 (Vote 102, April 22, 2026); at the Agriculture and Agri-Food committee (verified Aug. 1, 2026)First reading Sept. 18, 2025A live signal that food-price-transparency legislation can command a House majority in this Parliament — directly relevant to Toronto's Recommendation 4 collaboration ask
NDP unanimous-consent motion (Don Davies, backed by leader Avi Lewis)Defeated (unanimous consent denied)April 2026Would have had the House "agree that the government should ban surveillance pricing." (Attribution note, verified: the NDP's "This is unfair. It's a rip-off. And it's downright creepy" quote belongs to leader Avi Lewis, April 13, 2026 announcement — one capture tool misattributed it to a nonexistent "Bill C-266"; see Appendix I)
National AI Strategy ("AI for All")Released (policy document, not law)June 4, 2026Commits to strengthening privacy law "including against harmful practices such as... surveillance pricing"
National Food Security StrategyReleased (policy document, not law); $3.2B/10 yearsJune 11, 2026Recommends stronger privacy law and leveraging the Competition Act against algorithmic pricing/collusion
Competition Bureau discussion paper + "What We Heard"Consultation complete; no policy recommendations issuedPaper June 2025; report Jan. 22, 2026Surveys stakeholder views; explicitly not a Bureau policy position
Competition Bureau RealPage/Yardi rental-algorithm investigationDiscontinued, with compliance guidance issuedOpened Jan. 2025; closed Nov. 10, 2025Adjacent precedent (rental housing, not groceries) showing the Bureau's practical limits in proving algorithmic-pricing harm under current law
Competition Bureau food-supply-chain examinationOngoingAnnounced June 16, 2026; report due spring 2027Covers loyalty programs, algorithmic pricing, "shrinkflation"

6.2 Canada — provincial

JurisdictionInstrumentStatusKey date
ManitobaBill 49 — The Business Practices Amendment Act, S.M. 2026, c. 41Enacted and in force July 1, 2026 (proclamation signed June 17, 2026 — verified against the King's Printer proclamation record and consolidated C.C.S.M. c. B120, Aug. 1, 2026)Introduced Mar. 2026; Royal Assent June 1, 2026
OntarioBill 104 (Fair Grocery Prices Act, 2026; PMB, John Fraser, Ont. Liberal)First reading carried April 15, 2026, ordered for second reading; no further progress as of Aug. 1, 2026 (verified on ola.org; session live, bill parked, no scheduled date)First reading April 15, 2026
OntarioBill 113 (Fair Prices and Tax-Free Groceries Act; NDP)At second-reading stage per ola.org (Aug. 1, 2026) — note: contains nothing on algorithmic/surveillance pricing (HST + grocer property-covenant measures); confirmed the only other "grocery" bill in Parliament 44-1, so Bill 104 is Ontario's sole vehicle on point2026
OntarioConsumer Protection Act, 2023 (S.O. 2023, c. 23, Sched. 1; Bill 142, R.A. Dec. 6, 2023)Not yet proclaimed in force, no announced date (verified Aug. 1, 2026 — regulations consultation launched Dec. 2024). A capture-tool claim of an enacted "s. 74.0.1 algorithmic transparency" provision effective Jan. 1, 2027 is fabricated — apparently a garbled reference to Bill 104's proposed s. 9.1 (see Appendix I)2023–2026
OntarioLaw Commission of Ontario, Improving Consumer Protection in the Digital Marketplace: Final Report (32 CPA-amendment recommendations)Verified — published May 13, 2024. The wave's "July 16, 2026 LCO call" did not verify (no LCO news item exists between May 28, 2026 and Aug. 1, 2026); the modernization-call framing traces to law-firm coverage of the 2024 reportMay 13, 2024
QuebecPrice-accuracy rules (OPC Politique d'exactitude des prix: register price higher than advertised → item free if ≤$15, or advertised price minus $15 if above)Mechanics verified against the OPC's own page (threshold raised from $10 to $15 in 2025, Bill 72 reforms; exclusions apply — clothing, unbarcoded items, minimum-priced milk/alcohol, tobacco). The wave's framing of it as a structural deterrent to individualized checkout increases remains analysis, not an established findingUpdated Dec. 2025
YukonResidential Tenancies Act, SY 2025, c. 7 (replaced the former Residential Landlord and Tenant Act) — makes setting rent by algorithm an offenceEnacted and in force September 1, 2025 (verified against yukon.ca) — the first Canadian jurisdiction to make algorithmic rent-setting an offence; adjacent domain (housing, not groceries) but a Canadian first worth namingIn force Sept. 1, 2025
OntarioNon-binding NDP motion (Marit Stiles)Symbolic only, no legal effect2026
QuebecCAI decision blocking Metro's facial-recognition pilotFinal regulatory decision (privacy, not pricing)Feb. 18, 2025
QuebecLaw 25 (private-sector privacy law)In forceFull force since Sept. 2023
Saskatchewan, Nova ScotiaReported political advocacy only, per secondary sourcing (Policy Options)No bill text located—
BC, AlbertaNo engagement found——

6.3 United States — federal and state

JurisdictionInstrumentStatusKey date
Federal (FTC)6(b) study + "Issue Spotlight"Non-binding study; no rule, no enforcement actionStudy 2024; report Jan. 2025
Federal (Congress)3 distinct bills — numbering resolved this review: One Fair Price Act of 2025 (S.3387, Gallego, Dec. 2025); Stop AI Price Gouging and Wage Fixing Act of 2025 (H.R.4640); Stop Price Gouging in Grocery Stores Act (S.3892, Feb. 2026)Referred to committee, none reported out2025–2026
MarylandHB 895 (Protection From Predatory Pricing Act), 2026 Md. Laws ch. 154 — cross-file SB 387 diedEnacted, in force Oct. 1, 2026Signed April 28, 2026
ConnecticutPublic Act No. 26-64 (sSB 4 — same law)Enacted; § 11 pricing provisions AND §§ 1–10 data-broker regime all effective October 1, 2026 (corrected against the enrolled act — secondary reports of July 1, 2027 were wrong); AG-only CUTPA enforcement, no private rightSigned May 27, 2026
New YorkGen. Bus. Law §349-a (disclosure)Enacted and in force; First Amendment challenge dismissed at first instance (NRF v. James, S.D.N.Y., Rakoff J., Oct. 8, 2025); appeal pending at Second Circuit (verified Aug. 1, 2026)Effective July 8, 2025 (AG filing in NRF v. James; corrected on outside review from the secondary-source 'Nov. 10, 2025')
New YorkGen. Bus. Law §340-b (algorithmic rent-coordination ban — not groceries; listed to prevent recurring confusion)Signed Oct. 16, 2025 (S.7882); effective Dec. 15, 2025; enforcement stayed as to RealPage and customers pending RealPage v. James (S.D.N.Y.) preliminary-injunction ruling (verified Aug. 1, 2026)Effective Dec. 15, 2025
New YorkOne Fair Price Act (S.8623-B/A.9349-B)Passed both chambers June 2026; not yet delivered to the Governor — no signing clock running (the reported "Dec. 31, 2026 deadline" is not a statutory fact; verified Aug. 1, 2026)Passed June 2026
New JerseyFair Price Protection Act (ACS for A4085/A4523; Senate companion S3612)Enacted; ban in force Aug. 1, 2027; ESL moratorium in force Feb. 1, 2027; CFA enforcement route (verified against §9 commencement clause)Signed July 23, 2026
ColoradoHB 26-1210VetoedVetoed June 2, 2026
CaliforniaAB 2564Passed Assembly; Senate Privacy 5–2 (June 22, 2026); in Senate Judiciary (verified Aug. 1, 2026)Passed Assembly June 4, 2026
CaliforniaAB 325 (algorithmic-collusion antitrust rule, distinct from pricing ban)In forceSince Jan. 1, 2026
ColoradoSB 24-205 (AI Act, 2024), delayed by SB25B-004 (2025), then repealed and reenacted by SB26-189 (signed May 14, 2026)Enacted; obligations from Jan. 1, 2027, narrowed to notice/disclosure, data-correction and human-review rights; AG-only enforcement (verified)2024–2026
PennsylvaniaHB 1779 (Rep. Guzman) — mandates the NY-style "THIS PRICE WAS SET BY AN ALGORITHM USING YOUR PERSONAL DATA" disclosure and bars protected-class data in pricingProposed — in House Consumer Protection, Technology & Utilities Committee since Aug. 4, 2025 (verified on palegis.us)Introduced July 31, 2025
New York CityLocal Law 144 of 2021 — bias audits + notices for automated employment decision tools; DCWP enforcement since July 5, 2023Enacted and in force — the leading municipal algorithmic-accountability precedent (employment, not pricing)2021–2023

Over 40 surveillance-pricing-related bills were introduced across at least 24 U.S. states in 2026 by one advocacy tracker's count — more than triple the 2025 total — though this aggregate figure comes from a secondary/advocacy source (stateofsurveillance.org) and was not independently cross-checked against a primary legislative database such as NCSL in this research pass.

6.4 European Union, United Kingdom, Australia, and international bodies

JurisdictionInstrumentStatusNote
ChinaPIPL Article 24Enacted and in force since Nov. 1, 2021 (verified against the DigiChina translation, Aug. 1, 2026)Prohibits "unreasonable differential treatment of individuals in trading conditions such as trade price" via automated decision-making — the world's earliest major-economy statutory provision on point; no located enforcement action specifically on price discrimination
EUGDPR Article 22In force since 2018Contested applicability to ordinary price personalization (academic "sceptical view")
EUPrice Indication Directive Art. 6a (Dir. 98/6/EC as amended by Dir. (EU) 2019/2161)In force — verified this review via the Commission's Art. 6a guidance: "The prior price means the lowest price applied by the trader during a period of time not shorter than 30 days prior to the application of the price reduction"A structurally different lever: reference-price integrity rather than data-practice regulation — the direct regulatory answer to the Zhou discount-relabeling evasion (Ch. 5.3)
EUDigital Fairness ActProposed; Commission proposal expected Q3 2026Would address "unfair personalisation and profiling, including personalised pricing" and dark patterns
UKDigital Markets, Competition and Consumers Act 2024Enacted and in force (phased commencement: digital-markets provisions Jan. 1, 2025; consumer provisions during 2025 — exact commencement instruments not pinned in this research)First DMCCA consumer-enforcement drive launched Nov. 18, 2025 (verified on gov.uk): 8 investigations (StubHub, viagogo, Wayfair, others), advisory letters to 100 businesses across 14 sectors, and CMA209 price-transparency guidance — aimed at drip pricing and pressure selling, not personalized pricing (an earlier tool framing of a "dynamic-pricing review" was imprecise)
AustraliaPrivacy Act amendments (APP 1.7–1.9, automated-decision transparency)Enacted, in force Dec. 10, 2026General transparency rule, not pricing-specific
AustraliaACCC Digital Platform Services InquiryConcluded, recommendations onlyPublished June 2025; recommended (not enacted) economy-wide unfair-trading-practices rules
G7/OECD"Algorithmic Pricing and Competition in G7 Jurisdictions"Non-binding comparative researchPublished Oct. 2025, prepared for the Canadian-chaired G7 competition summit

Overall pattern across every jurisdiction surveyed: political attention, motions, consultations, and proposed bills significantly outrun enacted, in-force law. Manitoba is the only Canadian jurisdiction with an enacted statute — and as of July 1, 2026 it is proclaimed and in force, making it Canada's first operating surveillance-pricing law rather than a pending one. In the United States — the most legislatively active jurisdiction found — only three states (Maryland, Connecticut, New Jersey) have fully enacted grocery/retail-sector bans, one state (New York) has a narrower disclosure law actually in force, one governor (Colorado's) has vetoed a bill, and the largest state (California) has not yet passed its bill through even one full chamber's opposite house. No government anywhere in this survey has enacted a comprehensive, sector-general, symmetric (covering both increases and decreases) full prohibition — every enacted law found is either disclosure-only, asymmetric, or sector-limited to food/grocery retail.

Chapter 7 — The Equity and Vulnerable-Consumer Overlay

7.1 Who is actually harmed, and by what evidence

No Canadian study identifies which specific demographic groups are harmed by surveillance pricing in Canadian grocery, because — per Chapter 1 — no Canadian instance of the practice has yet been documented. The equity case therefore has to be built from three kinds of evidence: (a) documented harms in directly analogous sectors and markets, (b) structural reasoning about who disclosure-based remedies fail, and (c) Canada-specific data on who is most economically exposed to any grocery price increase generally.

Analogous documented harms:

Structural reasoning — who disclosure fails:

A disclosure-only remedy (Rung 1 of the Chapter 4 ladder) presumes the consumer can read the disclosure, understand it, and act on it — by comparison shopping elsewhere, refusing consent, or complaining to a regulator. Each of those steps has a real-world access cost that falls unevenly:

The two-directional vulnerability mechanism — the wave's most sophisticated equity contribution. (a structural analysis this report adopts and owns — the idea originated with two capture-wave submissions, and as stated here it rests entirely on facts already verified elsewhere in this report, making no new empirical claim): digital exclusion cuts both ways simultaneously. Shoppers without a smartphone, app, or loyalty account are structurally excluded from the personalized-discount side of these systems (app-only deals, member pricing, targeted coupons) even while electronic shelf labels newly expose them to whatever in-store price changes the system produces — they get the surveillance without the subsidies. App-enrolled shoppers face the reverse exposure: they receive the discounts but supply the behavioural data that willingness-to-pay models consume. A policy that only bans data-driven increases protects the second group; only ESL/point-of-sale transparency and baseline-price integrity protect the first. This mechanism explains why the ladder's cheap rungs (disclosure, ESL transparency) and its equity-driven design requirements below are not alternatives but complements.

Canada-specific economic exposure data (documented): the population most exposed to any grocery price increase, algorithmic or otherwise, is large and growing: 9.8 million people (24.0% of the population), including 2.4 million children, lived in a food-insecure household in Canada in 2025 (PROOF, University of Toronto, 2026); average family food spending is projected to rise by up to $994.63 in 2026 alone (Dalhousie, Canada's Food Price Report 2026). These figures do not distinguish algorithmic from ordinary price increases, but they establish the baseline vulnerability of the population any surveillance-pricing harm would land on hardest.

7.2 What a genuinely protective policy needs to do differently

Read together, the evidence above points to specific design requirements a disclosure-only or opt-out-only policy will not meet on its own:

  1. Default protection, not opt-out burden. An opt-out regime that requires the consumer to take action to avoid a higher price inverts the protection for exactly the consumers least able to take that action — plain-language disclosure paired with an opt-in consent requirement for any price increase (Manitoba's s.2(5) two-part test, Chapter 4 Rung 2) is a meaningfully stronger default than a bare disclosure label, because it puts the burden of action on the retailer wanting to charge more, not on the consumer wanting to avoid it.
  2. Multilingual, plain-language disclosure, tested with newcomer and low-literacy communities before adoption — none of the enacted disclosure laws surveyed (New York, Connecticut) specify a multilingual requirement in the text found in this research; Toronto could differentiate its own rule here.
  3. Point-of-sale, not app-only, remedies. Any rule that depends on a smartphone app to disclose or allow opt-out will systematically miss shoppers without one; electronic-shelf-label transparency (Chapter 4, Rung 8) has the advantage of being visible in-store regardless of a consumer's device access.
  4. Enforcement that does not depend on individual complaint or litigation. Private-recourse models (New Jersey, through its Consumer Fraud Act — New York's pending bill is AG-only) put real teeth behind a law, but shift enforcement cost onto consumers least likely to have the time, legal literacy, or resources to sue — reinforcing the case for pairing any private right of action with active regulator-driven enforcement (as Manitoba's Director of Business Practices model does) rather than relying on litigation alone.
  5. Outcome auditing, not just input-based rules. Because the insurance "bluelining" evidence shows algorithms can reproduce protected-characteristic discrimination without ever using a protected characteristic as an explicit input, a policy that only prohibits listed data categories (as Manitoba's and Ontario's near-identical definitions do) risks leaving a structurally identical harm untouched if achieved through a proxy variable. A more durable rule would pair the input-based prohibition with a periodic outcome audit requirement — no jurisdiction surveyed in Chapter 6 has yet adopted this, making it a genuine opportunity for Toronto/Ontario to lead rather than follow.

Chapter 8 — The Industry and Free-Market Counter-Case (Steelmanned)

This chapter presents the strongest version of the case against regulating or banning surveillance pricing, as made by its most substantive critics, without editorializing against it — readers should weigh it against Chapters 1, 3, and 7 themselves.

8.1 The Information Technology and Innovation Foundation (ITIF)

ITIF, a technology-policy think tank, has published the most sustained and specific critique found in this research, directly addressing Manitoba's Bill 49 and the Canadian federal NDP motion: Lawrence Zhang, "A Ban on Personalized Pricing Is Not Consumer Protection," ITIF, June 8, 2026 (itif.org). Its core arguments:

  1. A ban does not lower prices; it redistributes them. Faced with a uniform-price rule, a firm sets one price across its customer base that lands somewhere in the middle of the old price distribution — the top of the range (what high-willingness-to-pay customers previously paid) comes down, but the bottom of the range (what price-sensitive customers previously paid) goes up.
  2. Manitoba's asymmetric design (banning only data-driven price increases, not decreases) may not work as intended. ITIF argues firms facing this rule will either abandon personalization altogether (netting the same effect as (1) above) or simply reset their "standard" reference price upward and reframe everything below it as a discount.
  3. Empirical claim: over 60% of consumers may pay less under personalized pricing. The underlying study is now verified as peer-reviewed and top-tier: Jean-Pierre Dubé & Sanjog Misra, "Personalized Pricing and Consumer Welfare," Journal of Political Economy 131(1) (2023): 131–189 (doi:10.1086/720793; SSRN 2992257 — an earlier tool-supplied SSRN ID, 3035110, was wrong). The published abstract states: "While total consumer surplus declines under personalized pricing, over 60% of consumers benefit from personalization" — "benefit" meaning better off than under the optimized uniform price. This is the load-bearing empirical claim in the entire counter-case, and it directly complicates the equity narrative in Chapter 7 — some of the same price-sensitive shoppers a ban is meant to protect might be ones a ban makes worse off, by eliminating discounts they currently receive rather than price increases they currently suffer. Note the full sentence cuts both ways and should be quoted whole: total consumer surplus declines even as a majority gains. This report has verified the publication and finding but has not re-derived the methodology.
  4. Definitional overbreadth risk. ITIF argues both Manitoba's and the federal NDP's proposed definitions are broad enough to sweep in loyalty programs, member pricing, targeted coupons, and address-based delivery-fee variation — "longstanding features of market economies, not novel harms produced by Big Tech." New York's pending One Fair Price Act (Chapter 4, Rung 7) directly answers this specific objection by explicitly carving out loyalty, coupon, and senior pricing — evidence the objection is taken seriously even by drafters who ultimately still support a ban.
  5. Efficiency case study: dynamic perishable markdowns reduce food waste. ITIF cites a study (SSRN 4687311) finding an online grocer's automatic perishable-markdown personalization system reduced food waste in treated categories by 9.6%, arguing an overbroad ban could eliminate this kind of beneficial dynamic pricing along with the harmful kind.
  6. Existing law already covers the genuine harms. Collusion is illegal under the Competition Act; discrimination based on protected characteristics is illegal under human-rights law; abuse of dominance is illegal under the Competition Act. ITIF's position is that further intervention should be "proportionate and narrowly tailored" — transparency requirements, privacy protections, and clear anti-discrimination enforcement, not a new categorical ban.
  7. ITIF has made substantively the same argument in the U.S. state context ("State Data-Driven Pricing Bans Would Backfire on Consumers," ITIF, June 14, 2025) and submitted formal comments to the Competition Bureau's own consultation (Aug. 8, 2025), indicating an organized, cross-jurisdictional policy-advocacy position rather than a one-off commentary — relevant context for readers assessing the source, since ITIF is a technology-industry-adjacent think tank, not a neutral academic body, though its specific empirical citations above are independently traceable.

8.2 A more measured economic voice: Analysis Group

Rebecca Kirk Fair and Juan Carvajal, "The Rise of Surveillance Pricing," Analysis Group, 2025, published in the CPI Antitrust Chronicle (analysisgroup.com). This piece is notably more balanced than ITIF's advocacy framing: it acknowledges that price differentiation based on willingness to pay "can improve firm efficiency and consumer access to goods and services" while explicitly discussing "regulatory concern" about disparate impact and whether consumers can meaningfully consent to data collection they may not understand. This report could not independently verify whether Analysis Group's report was industry-commissioned; Analysis Group is a litigation-and-economic-consulting firm that performs paid client work generally, but no specific sponsor was identified for this particular publication in this research pass — flagged as an open question rather than an assumed conflict.

8.3 General free-market arguments, synthesized

8.4 Additions from the capture wave (verification status as marked)

This report's assessment of the counter-case: the strongest elements — the redistribution mechanism, the loyalty-program overbreadth risk, and the perishable-markdown efficiency case — are specific, testable, and at least partially supported by cited empirical work, and should be treated seriously rather than dismissed as industry talking points. The weakest element is the implicit suggestion that no intervention is warranted at all: even ITIF's own position (transparency requirements, privacy protections, clear anti-discrimination enforcement) is itself a rung on the ladder in Chapter 4 (Rungs 1–3), not a case for the pure status quo — meaning even the most skeptical serious critique found in this research converges with Toronto's own Recommendation 2(a) on disclosure, even as it diverges sharply from Recommendation 2(b) on prohibition.

Chapter 9 — What the Status Quo Costs, Versus What Each Option Costs to Implement and Enforce

No Toronto-specific costing exists yet — that analysis is precisely what staff's Q1 2027 report is supposed to produce, and this report does not have access to City budget or licensing-enforcement data that would allow a real dollar estimate. What follows is an order-of-magnitude framework built from the evidence gathered above, intended to structure (not substitute for) staff's own costing work.

9.1 The cost of the status quo

9.2 Approximate cost ordering by rung (implementation + enforcement, relative not absolute)

  1. Disclosure-only (Rung 1): lowest cost. For retailers, largely a UI/label change; no compliance-cost challenge to New York's disclosure law (effective July 8, 2025) was located in this research — an absence-of-evidence observation, bounded by the sources actually reviewed. For the City, enforceable via existing Municipal Licensing and Standards inspection/licensing processes at marginal additional cost.
  2. Disclosure + consent standard (Rung 2): low-moderate cost. Requires retailers to build a genuine point-of-sale consent flow (not just a static label), a meaningfully bigger technical lift than Rung 1, but still within existing e-commerce/POS system capability.
  3. Opt-out / predatory-data-practice restrictions (Rung 3 ask, Toronto's Rec. 2(c)): moderate cost. Requires building and maintaining an actual opt-out mechanism and associated complaint-handling capacity; likely needs dedicated enforcement staff rather than piggybacking on general licensing inspection, based on Manitoba's model of designating an existing Director of Business Practices office rather than building an entirely new one.
  4. Asymmetric/partial prohibition (Rung 4, Manitoba/Ontario Bill 104 model): moderate-high cost. Requires an enforcement body capable of auditing pricing algorithms and data practices (a genuinely technical function most municipal licensing offices do not currently have in-house), plus a formal complaints/investigation/penalty process; Manitoba's fine structure (up to $1M for repeat offences) suggests a serious, resourced enforcement office is expected to sit behind the law.
  5. Sector-specific full prohibition (Rung 5, Maryland model): high cost, but Maryland's own experience (Chapter 4) shows a poorly resourced, complaint-driven, AG-only enforcement model with a mandatory cure period and low penalty caps ($10,000–$25,000) risks being cheap to run but largely symbolic in effect — a cautionary data point on the gap between nominal and real enforcement cost/capacity.
  6. General hybrid prohibition + disclosure fallback (Rung 6, Connecticut model): high cost, but potentially more cost-effective per unit of actual protection than Maryland's narrower model, since the disclosure fallback catches conduct that would otherwise evade a narrowly drawn prohibition — worth flagging to staff as possibly the most cost-efficient design if Toronto or Ontario legislate at this level.
  7. Full prohibition with private right of action (Rung 7, New Jersey model): highest direct government cost avoided (litigation cost shifts to private parties and courts) but highest total systemic cost (litigation is expensive for everyone involved, including defendants who may ultimately be found compliant, and access-to-justice gaps mean the mechanism may go underused by exactly the vulnerable consumers discussed in Chapter 7 unless paired with active regulator enforcement as well).
  8. ESL-specific transparency/moratorium (Rung 8): low-moderate cost, and the most severable/cheapest meaningful action available — a moratorium on new ESL installations (New Jersey's model) costs the enforcing government very little (it is a permitting/registration function, not an ongoing audit function) while directly buying time for the broader legal and technical questions to resolve.

9.3 The bottom line for staff's costing exercise

The cheapest rungs (disclosure, ESL transparency) are also the ones Chapter 3 finds legally safest for Toronto to implement unilaterally — a fortunate alignment. The rungs that would most directly address the equity concerns in Chapter 7 (opt-in consent defaults, outcome auditing, non-app-dependent remedies) are also the most expensive and the ones most clearly requiring provincial or federal legislative backing to be enforceable against the full range of retailers and platforms operating in Toronto. This suggests a two-track resourcing strategy for the Q1 2027 report: budget for a real (not merely symbolic) municipal licensing-based disclosure/ESL-transparency regime now, while treating the prohibition rungs as an advocacy and model-legislation-drafting exercise directed at Queen's Park and Ottawa rather than a City budget line item.

Chapter 10 — Best-Practice Cases

Ranked here not by how aggressive each model is, but by how well-evidenced, well-targeted, and durable each appears to be — the criteria most relevant to a City drafting a report meant to survive legal and political scrutiny.

  1. New Jersey's Fair Price Protection Act (A4085/S3612, signed July 23, 2026) — the single best-designed enacted model found in this research. It combines a real prohibition (effective Aug. 1, 2027), consumer recourse through the Consumer Fraud Act's existing private action and treble damages (violations are CFA "unlawful practices" — direct consumer recourse, though via the CFA rather than a standalone new right), a new AG civil action (greater of actual damages or $50,000) in parallel, and a moratorium on new ESL installations from Feb. 1, 2027 pending a state impact study (a genuinely novel, evidence-gathering-first approach to the ESL-transparency question that neither Manitoba nor Toronto's own motion has adopted). (New Jersey Monitor, July 23, 2026; EPIC, July 23, 2026)
  2. Connecticut's Public Act 26-64 — the best model for exactly the graduated, one-statute ladder Toronto's Recommendation 2 asks staff to design: a real prohibition for the clearest cases, with a disclosure fallback ("THIS PRICE WAS INCREASED BY A PRICE SETTING DEVICE USING YOUR PERSONAL DATA") for conduct that falls outside the prohibition's scope, applied to retail generally rather than food alone. (EPIC, June 4, 2026) Now in force October 1, 2026 — the earliest-operating general-retail regime surveyed.
  3. New York's pending One Fair Price Act — the best model for avoiding the definitional-overbreadth objection raised by industry critics (Chapter 8): its explicit statutory carve-outs for loyalty programs, coupons, and senior pricing directly answer the strongest specific critique in this report while still reaching the core harm. Its already-in-force sibling law (Gen. Bus. Law §349-a) is additionally the best-tested model for surviving a First Amendment/compelled-speech challenge, relevant to any Canadian Charter-based free-expression challenge a Toronto or Ontario disclosure mandate might eventually face. (Wilson Sonsini; Skadden, Jan. 20, 2026)
  4. Manitoba's Bill 49 consent standard (s.2(5)) — the best model specifically for the equity concern in Chapter 7: its two-part test (prominent, clear disclosure of the reason for a higher price, plus overt affirmative consumer action) is a meaningfully higher bar than a passive on-screen label, and is the closest enacted model to the "default protection, not opt-out burden" principle this report recommends in Chapter 7.2. (web2.gov.mb.ca/bills/43-3/b049e.php)
  5. The Competition Bureau of Canada's "What We Heard" consultation process itself — a best-practice model for how a report like Toronto's should be built: broad public consultation (103 responses across individuals, businesses, academics, legal community, and consumer groups), an explicit, unusual candour about the limits of its own findings ("this report... does not necessarily reflect the Bureau's views"), and a clear-eyed acknowledgment of what the underlying law can and cannot currently reach. (Competition Bureau, Jan. 22, 2026)
  6. EPIC's ongoing issue-tracking page — a best-practice model for how Toronto's own staff could monitor this fast-moving field going into the Q1 2027 report; EPIC's page functions as a running, dated, sourced log of enactments, vetoes, and legislative testimony across US jurisdictions, updated as events occur rather than as a static one-time survey. (epic.org/issue-types/surveillance-pricing/)
  7. The OECD's "Algorithmic Pricing and Competition in G7 Jurisdictions" report — the best available international comparative-research model, prepared for the 2025 G7 competition summit under Canada's own G7 presidency; useful precisely because it stays disciplined about scope (explicitly competition-law-focused, flagging consumer-protection and privacy angles as adjacent but out of scope) — a model of the kind of scope discipline a Toronto staff report juggling municipal, provincial, and federal levers simultaneously will also need. (OECD, Oct. 2025)
  8. Quebec's CAI enforcement action against Metro's facial-recognition pilot — not a pricing case, but the best Canadian precedent for a privacy regulator actually stopping a specific grocery-retailer data practice before deployment, on a real statutory consent standard (the Québec IT Act's s.44/45 express-consent requirement), rather than after-the-fact. A useful model for what a genuinely proactive (not just reactive/disclosure-based) privacy-regulator role could look like if applied to pricing algorithms specifically — relevant to Toronto's Recommendation 7 request for OPC advice. (Osler, 2025)

Answering the Council's Seven Directives, in Summary

(1) All possible mechanisms to ban/regulate surveillance pricing raising grocery prices: the full ladder in Chapter 4 — disclosure, disclosure-plus-consent, general ADM transparency, asymmetric partial prohibition, sector-specific full prohibition, general hybrid prohibition-plus-disclosure, full prohibition with private right of action, and ESL-specific transparency — with a real, named, enacted-or-pending model at every rung.

(2) Jurisdictional scan/analysis from disclosure to full prohibition, opt-out, and ESL transparency: Chapter 3 (authority), Chapter 4 (the ladder itself), and Chapter 5 (who can actually enforce what against whom) together constitute this scan; this report's assessment is that disclosure and ESL-transparency measures are municipally defensible today, while opt-out and prohibition measures likely require provincial or federal legislative backing to be durable and broadly enforceable (Chapter 3.5, 5.3).

(3) Should Ontario ban surveillance pricing and strengthen the Consumer Protection Act: Chapter 6.2 and Chapter 3.3 — Ontario already has a ready-made vehicle (Bill 104, whose operative definition is near-verbatim the Manitoba statute that is now in force), carried at first reading April 15, 2026 and parked awaiting second reading with no scheduled date, while the sitting government has publicly rejected the idea; On the question as asked — should Ontario act — this report's assessment, on the evidence assembled here, is yes, at least to the disclosure-and-consent tier: Ontario is the level with undisputed jurisdiction, an in-force model one province west, a drafted vehicle (Bill 104) on its own order paper, and enforcement machinery (the CPA regime) a municipality lacks; the strongest counter-arguments this research surfaced (Chapter 8) go to how such a rule is drafted — carve-outs, baseline-price integrity, evasion-proofing — not to whether the provincial level is the right one. That assessment is offered as analysis for Council's judgment, not a substitute for it. The contrast sentence now available to Council, fully verified: Manitoba has enacted and brought into force what Ontario has only proposed.

(4) Collaboration with the federal Food Security Strategy and AI Strategy: Chapter 6.1 — both strategies already name surveillance pricing as a harm to be addressed through strengthened federal privacy law, giving Toronto's collaboration ask a genuine, receptive federal policy hook to attach to, principally through Bill C-36.

(5) Advice from ISED on PIPEDA: Chapter 6.1 and Chapter 5.3 — current PIPEDA is silent on surveillance pricing; Bill C-36 (sponsored by the Minister of Artificial Intelligence and Digital Innovation) is the live legislative vehicle — first reading completed June 15, 2026, awaiting second-reading debate — that would replace PIPEDA and create a new dedicated enforcement commission. Its verified definitional hook (s. 2(1): personal information "includ[es] information that is inferred about the individual"; s. 2(2): de-identification does not launder personal information out of scope) is the precise language ISED advice to Toronto should be asked to interpret against pricing-inference models.

(6) Advice from the Competition Bureau: Chapter 5.3 and Chapter 6.1 — the Bureau's own stated position is that it does not regulate prices directly and can reach surveillance pricing only via collusion, predatory-pricing, or deceptive-marketing theories under the existing Competition Act; its RealPage/Yardi investigation (discontinued, guidance issued) is the closest real precedent for how far those theories currently reach in an adjacent Canadian market.

(7) Advice from the Office of the Privacy Commissioner of Canada: Chapter 6.1 and Chapter 5.1 — the OPC's current enforcement role over private-sector data practices would transfer to a new Digital Safety and Data Protection Commission under Bill C-36; the OPC's own 2020 PIPEDA-reform recommendations (rights around automated decision-making, explanation, and objection) remain the most relevant existing OPC policy position, though none was issued specifically on surveillance pricing.

Recommendations for the Q1 2027 Staff Report

These follow directly from the analysis above and are offered as a starting point for staff's own judgment, not a substitute for it.

  1. State plainly that surveillance pricing is not yet documented in Canadian grocery retail, while explaining why the underlying infrastructure and commercial incentive make it a legitimate precautionary target — the credibility of the entire report depends on getting this distinction right at the outset (Chapter 1.2).
  2. Pursue disclosure and ESL-transparency measures now, via business-licensing conditions, as the legally safest and cheapest near-term action within confirmed municipal authority (Chapters 3.5, 4 Rungs 1 & 8, 9.3).
  3. Direct the prohibition-level asks at Queen's Park using Ontario's own Bill 104 as the ready-made vehicle, rather than attempting a City-wide prohibition of doubtful legal authority (Chapters 3.5, 4 Rung 4, 6.2).
  4. Model any Toronto- or Ontario-drafted consent standard on Manitoba's two-part test (s.2(5)) rather than a passive disclosure label alone, to better serve the vulnerable-consumer population identified in Chapter 7.
  5. Explicitly carve out loyalty programs, coupons, and senior/student discounts, following New York's pending model, to pre-empt the strongest and most specific industry objection (Chapter 8.1, point 4) and avoid Maryland's documented drafting gaps (Chapter 4, Rung 5).
  6. Flag the cross-border enforcement gap honestly — but state the doctrine accurately, because it has moved. Do not assume any City or provincial rule will reach out-of-jurisdiction online retailers, delivery platforms, and data brokers; but equally, do not overstate the gap: the verified Libman→Sharp→Clearview (2026 BCCA 67) line establishes that provincial regulators can reach foreign actors whose data collection connects them to the province. Route the residual (municipal-reach and practical-collection) problems toward the federal Bill C-36 process and Competition Bureau/OPC advice already requested under Recommendations 5–7 (Chapters 3.6, 5.3).
  7. Consider an outcome-auditing requirement alongside any input-based prohibition, to guard against the proxy-variable evasion risk documented in the insurance "bluelining" literature (Chapter 7.2) — an area where Toronto could set a genuinely novel best practice rather than simply importing an existing model.

Appendix I — Integration record: what was verified, what was folded in marked, what was checked and rejected (August 1, 2026)

Method: three same-day verification passes totalling nine parallel primary-source lanes (pass 1: statutes/case law, legislative status, secondary harvest, plus a triage lane over all 19 extraction packets; pass 2: Canadian law & Toronto record, US/EU statutory conflicts + fresh-status sweep, empirical/industry claims; pass 3: browser-rendered gates (RCC letter, Connecticut enrolled act), fold-in candidates, and the s. 11 conflict-doctrine base). No new multi-AI capture was run; the targeted Turn-3 follow-up (RUBRIC_SCORES finding #1) is now prepared as an operator dispatch at this library's internal records and awaits manual execution. The master briefing (sonnet-base/one of this library's internal records), SET-INDEX (one of this library's internal records), one of this library's internal records, and one of this library's internal records are all preserved unchanged; this v1.0 supersedes both prior drafts.

I.1 Verified this review (primary source touched; safe to treat as settled subject to the dates given)

#ClaimPrimary sourceVerdict
1City of Toronto Act, 2006, ss. 6(1), 6(2), 7, 8(2) (paras 5, 8, 11), 8(3), 10, 11 — text and effectontario.ca/laws/statute/06c11Verified — "consumer protection" is express; yuanbao's find is real (its "(h)/(k)" lettering corrected to numbered paras 8/11; its s. 6(2) framing overstated — s. 6(1) does the work)
2Manitoba Bill 49 statusKing's Printer of Manitoba: S.M. 2026, c. 41 chapter page; proclamation of 2026-07-01; consolidated C.C.S.M. c. B120Verified — enacted AND IN FORCE July 1, 2026 (assent June 1, 2026; proclamation signed June 17; publication date unconfirmed — outside review). Corrects the master briefing ("not yet proclaimed" — stale after June 17) and the master briefing's assent date (was recorded as May 5)
3Ontario Bill 104 (Fair Grocery Prices Act, 2026)ola.org bill page + full 44-1 bill sweepVerified — Fraser (Ont. Lib.), first reading carried Apr. 15, 2026, ordered for second reading, no progress; only Ontario bill on point; competing tool-supplied bill numbers/sponsors wrong
4Libman v. The Queen, [1985] 2 S.C.R. 178, para. 74 testSCC decisions databaseVerified (existence, citation, test) — criminal-territoriality case; use as doctrine-origin only, with Sharp, 2023 SCC 29 as the modern regulatory authority
5Clearview AI Inc. v. B.C. (IPC), 2026 BCCA 67, aff'g 2024 BCSC 2311Appellate record + multiple firm summariesVerified — real-and-substantial-connection holding against a foreign platform; qwen's court+date accurate, neutral citations added
6Coun. Dianne Saxe quote ("It's a stretch...the abuse is real")TorontoToday.ca, Chamandy, July 30, 2026 (rendered live)Verified word-for-word; Effoduh's "modest things" confirmed in same article
7N.Y. GBL §349-a — content and litigation statusnysenate.gov consolidated text; NRF v. James recordVerified, then corrected on outside review — effective July 8, 2025 (AG filing in NRF v. James); the secondary-source 'Nov. 10, 2025' was wrong; Rakoff dismissal Oct. 8, 2025; 2d Cir. appeal pending
8N.Y. GBL §340-b — actual characternysenate.gov consolidated text; RealPage v. James docketVerified as a rental-housing algorithmic-coordination statute, enforcement stayed as to RealPage; the master briefing's reading was right
9Bill C-36 ss. 2(1), 2(2) definitional text + statusparl.ca first-reading text; LEGISinfoVerified verbatim ("information that is inferred about the individual"; de-identified info remains personal information); first reading June 15, 2026, no second-reading debate yet
10Bill C-226 (food price transparency)LEGISinfoVerified — second reading passed 168–150 (Vote 102, Apr. 22, 2026); at AGRI committee
11China PIPL Art. 24Stanford DigiChina translationVerified in substance (qianwen's wording is a faithful alternate translation); in force Nov. 1, 2021; no located price-discrimination enforcement action
12Toronto (City) v. Ontario (A.G.) delegated-powers propositionSCC recordVerified as 2021 SCC 34 — the proposition is good law; the tool-supplied citation was wrong (below)

Second pass (same day) — the I.3 queue burned down:

#ClaimPrimary sourceVerdict
13Croplife Canada v. Toronto2005 CanLII 15709 (ON CA), 75 O.R. (3d) 357Verified — but construed Municipal Act, 2001 s. 130 (pre-COTA); persuasive-analogous only
14Shark-fin by-law struckEng v. Toronto (City), 2012 ONSC 6818Verified — decided under COTA itself; closest precedent for a COTA by-law failing review
15Toronto food-insecurity emergency declaration2024.MM24.42 (adopted Dec. 17–18, 2024)Verified; the 24.7% figure is real but lives on the City's food-security page, not in the declaration (which cites 51% food-bank growth)
16MM39.27 algorithmic-pricing language2026.MM39.27, Decision 2(a)Verified — added by Chow amendment; a second live Council directive incl. algorithmic-pricing disclosure, Q2 2027
17Quebec price-accuracy policyOPC (opc.gouv.qc.ca), upd. Dec. 2025Verified — threshold is $15, not $10 (raised 2025, Bill 72)
18Ontario CPA 2023 statuse-Laws/ICLG/ministry recordVerified — R.A. Dec. 6, 2023, NOT in force, no announced date
19Maryland vehicle & mechanicsmgaleg.maryland.gov, enrolled hb0895e.pdfResolved — HB 895 alone (ch. 154); cross-file SB 387 died; "SB 597" wrong; no day-fixing rule; $10K/$25K are generic MCPA §13-410 penalties
20NJ dates & enforcementA4085 second reprint §9; Governor's releaseResolved — ban Aug. 1, 2027; ESL moratorium Feb. 1, 2027; CFA route (no standalone PRA); AG remedy ≥$50,000
21US federal bill numberingcongress.govResolved — OFPA = S.3387 (Gallego); H.R.4640 and S.3892 are different bills
22Connecticut identity & scopecga.ct.gov + consistent analysesResolved — sSB 4 = PA 26-64, signed May 27, 2026; data-broker registry same act, phased 2027–2031. Residual: internal section number/verbatim (enrolled PDF resisted automated fetch)
23Colorado veto letterGovernor's letter via Colorado Newsline/EPICVerified — overbreadth reasoning confirmed verbatim ("would punish differentially lower prices")
24EU Price Indication Directive Art. 6aCommission Art. 6a guidance (CELEX:52021XC1229(06))Verified — 30-day prior-lowest-price rule
25Dubé & MisraJPE 131(1) (2023) 131–189; SSRN 2992257Verified — ">60% of consumers benefit"; total surplus declines; tool-supplied SSRN 3035110 was wrong
26Stamatopoulos/Sanders/Bray ESL studySSRN 5271491Verified (working paper) — no surge-pricing change after ESL adoption
27Bureau "over 60" pricing-algorithm vendorsBureau discussion paper textVerified — with caveat: figure links to a Capterra directory count
28Eversight/SOLUM/Pricer vendor clusterCompany announcementsVerified with nuances — Instacart acquired Eversight Sept. 2022; SOLUM's dynamic-pricing module is marketed generally, NOT tied to Loblaw; Pricer/Sobeys 5M labels was an announced target (+US$51M deal Apr. 2026)
29Zhou discount-relabeling argumentYale Insights, June 1, 2026Verified — expert commentary, not peer-reviewed
30Cineplex penaltyBureau release; FCAVerified — $38.9M (Sept. 2024); FCA affirmed Jan. 21, 2026; SCC leave announced (no docket entry located)
31OPC Tim HortonsPIPEDA Findings #2022-001Verified (June 1, 2022)
32Polling granularityAbacus release; UFCW/GBAO memo 052626Verified — Abacus 13% prior awareness; UFCW survey is May 26, 2026, U.S. voters, union-commissioned (67% ban / 65% ESL / 68% SP price-increase expectations)
33Fresh statusnysenate.gov; CA committee recordsVerified — NY OFPA passed but not delivered, no signing clock; CA AB 2564 in Senate Judiciary; no new state enactment since July 23, 2026
34Canadian Shield Institute / Bednarcanadianshieldinstitute.ca; Globe and MailVerified — org exists (launched ~May 2026, explaining earlier dead URL); Bednar is its managing director; her C-36 quote is from Globe coverage, June 2026
35RCC on-record position + Furlong identityRCC letter filed on 2026.EX33.33 (communicationfile-218619.pdf); retailcouncil.org pages; lobbyist registryVerified verbatim — and the letter is on the Council record itself
36Connecticut PA 26-64 § 11Enrolled act PDF (rendered)Verified with corrections — § 11 confirmed; effective Oct. 1, 2026 (not July 1, 2027); disclosure string is the longer "PRICE SETTING DEVICE" formulation; exceptions list differs from NY's (no errors/outages); CUTPA AG-only
37NYC Local Law 144nyc.gov DCWPVerified — municipal AEDT bias-audit law, enforced since July 5, 2023
38Tiny Township STR precedent2026 ONCA 408, aff'g 2025 ONSC 1578Verified — by-law upheld under licensing power + community well-being
39Yukon algorithmic rent-setting offenceyukon.caVerified, corrected — Residential Tenancies Act, SY 2025, c. 7, in force Sept. 1, 2025 (not a 2026 RLTA amendment)
40UK CMA actiongov.uk, Nov. 18, 2025Partly verified, reframed — drip-pricing/price-transparency enforcement drive (8 investigations, letters to 100 businesses), not a "dynamic-pricing review"
41Colorado SB 24-205 supersessionleg.colorado.govVerified — SB26-189 (signed May 14, 2026) repealed and reenacted, obligations Jan. 1, 2027
42Pennsylvania HB 1779palegis.usVerified — NY-mirror disclosure bill, in committee since Aug. 4, 2025
43Solomon tabling framing + CCLA critique of C-36Globe and Mail; ccla.orgVerified — quote real; "iPolitics" attribution wrong; CCLA "superficially de-identified" verbatim confirmed
44Instacart average spreadGroundwork/CR report + data repoVerified — 13% average high–low spread per multi-priced item; ~7% average basket difference; 23% max
45s. 11 conflict-doctrine baseSCC database; ontariocourts.ca; e-Laws; Justice LawsVerified — Spraytech 2001 SCC 40 (paras 35, 37–38); Rothmans 2005 SCC 13 (paras 12, 15, 22–25); Croplife para. 74 (s. 14 test applied to a Toronto by-law, upheld); CWB 2007 SCC 22 (paras 74–75); CPA 2002/2023 and Competition Act contain zero algorithmic-pricing provisions; Competition Act s. 50 repealed 2009

I.2 Checked and rejected (do not use; recorded so the error does not recur)

Added by the second pass:

Added by the third pass:

I.3 Publication tripwires — re-check on the day of use

All prior queue items are resolved. What remains is not verification debt but live-event risk; re-check each on the day this report is published or filed: NY One Fair Price Act delivery to the Governor (no clock is running yet); National Retail Federation v. James at the Second Circuit; the RealPage v. James preliminary-injunction ruling (NY GBL § 340-b enforcement stay); Cineplex's SCC leave application; Bill C-36's second-reading debate; Connecticut PA 26-64's October 1, 2026 commencement; Maryland's October 1, 2026 commencement; and any first enforcement action under Manitoba's in-force law. Added on outside review (I.5): a browser recheck of the Saxe TorontoToday page (it 403'd to the reviewer's lane; this project's first-pass lane rendered it live); the New Jersey enrolled-text section pins (§§ 3, 4, 5, 7, 9 — corroborated, not primary-verified); CanLII verbatim checks for Eng, 2012 ONSC 6818 and Tiny, 2026 ONCA 408 (both 403'd to all lanes; substance corroborated via case commentary); the Competition Bureau '60+ vendors' footnote; and the SCC docket for any actual Cineplex leave filing. Deliberately not folded in (low value or out of scope, recorded so the decision is visible): South Korea PIPA (no pricing ban — see I.2), Walmart/Amazon pricing-patent claims, kimi's data-broker ecosystem figures, and the wave's invented Toronto cost estimates (I.2).

I.4 Publication status — Council-ready (v1.0)

Every load-bearing claim in this report has been verified against a primary source across three same-day passes (nine lanes; 45 claims register items verified, 16 tool claims rejected and recorded) and stress-tested by a second-family outside review under the project's symmetric review rule (Kimi k3-256k — verdict SHIP-WITH-FIXES; all 19 findings adjudicated and applied in v1.1, Appendix I.5). No unverified-commodity marks remain in the body. The report's standard limitations, stated plainly rather than hidden: Chapter 3's legal analysis (including 3.5A) is policy analysis, not legal advice — the City Solicitor's opinion controls for Council, exactly as it would for staff's own report; all statuses are current to August 1, 2026, subject to the I.3 tripwires; and the optional Turn-3 capture follow-up (this library's internal records) remains available as enrichment, not as a gate.

I.5 Outside review and adjudication (a recorded standing decision symmetric review, August 1, 2026)

Review of record: 2026-08-01_KIMI_K3_REVIEW_SHADOW_REPORT_v1.0.md (Kimi k3-256k, five isolated fetch lanes; verdict SHIP-WITH-FIXES — 0 blockers, 4 major, 9 minor, 6 nits; zero findings on fabrication, mark-leakage, or taxonomy discipline). Disposition of all 19 findings, adjudicated at the Claude seat:

The reviewer's zero-findings on fabrication and precautionary-drift, and its 11/12 verbatim quote-audit pass rate, are recorded as the strongest external evidence yet for the report's core discipline.

Sources

Every source cited inline above, consolidated and organized by topic. Status/date qualifiers are repeated here where they affect how a citation should be used. Sources whose full text could not be independently fetched in this research pass (relied on via search-result snippets or secondary aggregation only) are marked [secondary/unverified in full text]. Outside review flagged bare-homepage citations against THE PROMPT's exact-URL rule; deep URLs are substituted above wherever this project banked one, and every remaining non-deep citation is now explicitly marked "[deep URL not banked — flagged per source rules]" rather than left silently as a homepage.

A. Toronto council record (primary)

C. Canada — federal government, legislation, and regulators

D. Canada — provincial

E. United States — federal and state

F. European Union, United Kingdom, Australia, and international bodies

G. Documented practice in Canadian grocery retail

G.1 Platform-by-platform survey (added in response to a follow-up request; U.S.-headquartered and Canadian-owned delivery/grocery platforms operating in Toronto)

H. Equity, economic, and polling data

I. Industry and free-market counter-case

J. Advocacy trackers and comparative surveys

End of report. This is an independent research draft, not a legal opinion and not a City of Toronto document. Several cited items are flagged above as secondary or unverified in full text; anyone relying on this report for drafting or advocacy purposes should re-confirm those specific items against primary sources before quoting them as settled fact, and should re-check the "pending"-status items (Bill C-36, NY's One Fair Price Act, California's AB 2564, the EU Digital Fairness Act, Manitoba's Bill 49 proclamation date, Ontario's Bill 104) for status changes given how quickly this field is moving as of mid-2026.

K. Added in the August 1, 2026 integration pass (primary sources consulted by the verification lanes)

Second pass additions (August 1, 2026):

Third pass additions (August 1, 2026):

Outside-review pass (August 1, 2026):