Technology, AI & the Automated City

A plain-language explainer · 8th-grade reading level · every figure below is sourced and checkable.

What’s wrong here

Here is the honest starting point: nobody elected the software. City staff are already piloting computer programs that touch real decisions — Toronto Public Health is testing predictive AI that would use more than 20 years of restaurant inspection records, plus 311 calls and wastewater data, to forecast which businesses are likely to break food-safety rules and decide who gets inspected first.¹ The city’s own report on that proposal doesn’t describe a step where a person double-checks the computer’s call before an inspector acts, or a way for a flagged business to appeal. In 2025, the city wrote its own AI policy in-house, and — drawing on how the city of San José, California does it — set out to build Canada’s first municipal “algorithmic impact assessment”: a working group is now building an AI policy, an AI registry, and that assessment framework.² In plain terms, an algorithmic impact assessment is a safety check the city runs on a computer program before it helps decide something about you, to catch unfairness before it happens — the same idea as testing a bridge before you drive over it.

Then there’s the street outside your door. In April 2025, Ontario’s Ministry of Transportation approved a company called Magna to test self-driving delivery vehicles on Toronto’s own streets. The City of Toronto has no say over that pilot. None. The province only invited city staff to look it over and comment.³ It is Toronto’s pavement, Toronto’s sidewalks, Toronto’s pedestrians — and a decision made entirely in Queen’s Park.

And not everyone is even online to notice. Roughly 58,000 Toronto residents — about 2% of the city — still lack affordable home internet.⁴ As city services move onto apps and websites, whoever designed the “digital city” needs to remember the resident with no smartphone, no home Wi-Fi, or no English.

What’s been proven to work elsewhere

Toronto isn’t inventing this from scratch. The city is explicitly modeling its algorithmic impact assessment on San José’s process — a city that has already built and used one.² The European Union runs something similar for any organization, public or private: before using a computer system that could seriously affect people’s rights, it must complete a formal risk review — checking what the system does, whether it’s really necessary, what could go wrong, and how those risks will be reduced.⁵ That is the same basic idea Ontario is now applying to city and provincial bodies through law, not just guidance.

Ontario has also just written a real deadline into law. Amendments passed in 2026 (Bill 97) require public bodies like the City of Toronto to complete a “privacy impact assessment” — a check for how a new system will collect and protect personal information — before they start collecting that information, and to report data breaches to both the affected person and the province’s privacy watchdog. The rules take effect in stages, starting July 1, 2026 and January 1, 2027.⁶ That is the province building real teeth into the same idea Toronto’s own policy is reaching for.

What we could choose together

This is a category where jurisdiction really matters, and pretending otherwise would be dishonest. The city can build and enforce its own rules for how it uses AI on residents — the algorithmic impact assessment is a city-level project, already underway.² The city can also set standards for the outside companies it hires to handle resident data, since it is the one signing those contracts.

But the city cannot regulate what drives on its own streets. Only the province can. Ontario’s Automated Vehicle Pilot Program has been extended to October 2027⁷ — and every decision about which self-driving vehicles get approved for Toronto’s roads, like Magna’s, is made in Queen’s Park, not City Hall.³ If a candidate wants the city to have a real voice over automated vehicles on its own streets, that voice has to be won from the province — it doesn’t exist today.

There’s also the question of city workers themselves. Ontario’s 444 municipalities together employ more than 235,000 people — almost 4% of everyone working in the province.⁸ As automation reaches more city work over the next term, someone will decide which tasks get automated and which stay human — and right now, that decision doesn’t clearly belong to anyone.

None of this is about fearing computers or worshipping them. It’s about who writes the rules, who gets asked first, and who gets left out if nobody checks.

Sources & receipts

1. Toronto Public Health’s proposed predictive-AI modelling using 20+ years of DineSafe inspection data, 311 calls, and wastewater surveillance to forecast food-safety infractions and prioritize inspections; no human-review or appeal step documented in the city’s own report.
2. Toronto’s 2025 in-house AI policy, modeled on San José’s process; first municipal algorithmic impact assessment in Canada; an interdivisional AI Working Group building an AI policy, an AI registry, and the assessment framework.
3. April 2025: Ontario’s Ministry of Transportation approved Magna to test automated last-mile delivery vehicles on Toronto streets; the City of Toronto has no regulatory authority over the pilot, only an invitation to comment.
4. Roughly 58,000 Toronto residents (about 2%) lack affordable in-home internet — City of Toronto ConnectTO Program Update, July 2025.
5. EU General Data Protection Regulation, Article 35: a data protection impact assessment is required before high-risk data processing.
6. Ontario Bill 97 amendments to the Municipal Freedom of Information and Protection of Privacy Act: privacy impact assessments required before collecting personal information; breach reporting to the affected person and the province’s Information and Privacy Commissioner; phased in July 1, 2026 and January 1, 2027.
7. Ontario’s Automated Vehicle Pilot Program (O. Reg. 306/15) extended to October 13, 2027.
8. Ontario’s 444 municipalities employ more than 235,000 people, about 4% of Ontario’s paid workforce — Association of Municipalities of Ontario.

This explainer went through an independent verification pass before publication (fact-check against primary sources, claim by claim). The internal verification record has been removed from this public page; the sourced facts and figures above are unchanged by that removal.