THE PROBLEM — Policing and surveillance in Toronto

1 · The problem

Policing is Toronto's single biggest discretionary budget line: $1.43 billion net in 2026, up 7.0% year over year, funding 143 net new officers. Layered onto that budget is a growing surveillance apparatus — automated licence-plate readers already scan more than a million plates a day across Toronto Police Service vehicles, and any high-risk AI system the service wants to deploy, including facial recognition, needs the police board's approval first. That board-and-budget decision point is the whole lever: once a surveillance system is purchased and installed, it is almost never removed, so the real window to shape what policing and surveillance look like for the rest of this decade is the procurement decisions happening this council term, not some later review. Everyone living in the city is policed by this budget, but the burden lands unevenly — racialized communities carry documented disparities, and every resident's data footprint grows with each new system the board approves. The city controls the budget and the board, among the strongest levers anywhere in this register; the province sets standards through the Community Safety and Policing Act and the Inspector General of Policing; the federal government holds criminal law. Full framing and receipts: register entry; scheduling context: this term's AI and surveillance procurement decisions are also inventoried in DECISIONS BEFORE 2030 — the full inventory (Phase C, one of this library's own project records) §1–2.

2 · What we're asking — the question hierarchy

The primary question: What is Toronto actually buying with its largest discretionary budget line — police service and the surveillance technology layered onto it — and does that spending make anyone measurably safer?

What the budget buys, and who signs off

• Toronto's police budget funds both officers and a growing tier of surveillance technology — what does that spending downstream of scarcity (police call-outs, courts, incarceration, ER visits, shelter churn) actually cost, set against what closing the income floor upstream would cost instead, both columns built honestly? (mci MC49)
• Set against the cost of running the current system — shelter beds, hospital cycles, policing of public space, and encampment clearances — what would it actually cost to end homelessness instead of maintaining it, and why does the city keep choosing the more expensive option? (homes HM7)
• What is the complete, current inventory of TPS surveillance assets — plate readers, camera networks, any facial-recognition-capable system — and the board-approval date attached to each? (the register's own named gap)

The surveillance line — what a system may decide

• Some decisions a machine could make accurately, and we should still forbid it anyway — sentencing a person, deciding who gets the apartment, "the surveillance judgment" itself. Is there a charter-worthy "reserved to humans" list the board should adopt before approving its next high-risk system? (automata AU50)
• Valuation and decision systems already run quietly elsewhere — the credit score, the risk model, the algorithm that helps decide who gets the loan, the job, the apartment, the parole. Is that the same class of system the police board is being asked to pre-approve, and does a resident have the right to see the model that judged them, contest it, and demand a human instead? (automata AU86)
• Once the board approves a surveillance system, what review or sunset mechanism exists — given that once installed, surveillance is nearly never uninstalled, is pre-approval the only check that will ever apply to it?

Crime, need, and what actually reduces harm

• What share of Toronto's actual acquisitive crime — theft, shoplifting, break-ins — is driven by need or fear of not having enough, versus organized profit or status, measured rather than assumed? (mci MC43)
• The strongest evidence for a needs-floor effect sits at the level of the individual dependent user; the weakest link is population-wide — Toronto's own Housing First trial found no significant drop in arrests. What's the honest causal chain, and which part of the crime problem does a needs floor not fix on its own? (mci MC44)
• Some theft isn't about need at all — fraud tracks with wealth, organized retail crime is an industry. Where exactly is the honest boundary of the "poverty drives crime" claim, stated before anyone leans on it? (mci MC48)
• What would a "no one prosecuted for poverty" audit find in Toronto today — which bylaws, enforcement patterns, and fare-inspection practices functionally criminalize being poor? (mci MC51)
• The strongest case against a needs-floor approach: scarcity is real, but crime is still a choice, and subsidizing dependency can entrench it. Toronto's own policy debate deserves that argument made at full strength, then answered with evidence, not dismissed. (mci MC76)
• Skateparks and parkour spots are where kids are most often chased off and ticketed for informal, self-taught recreation — what would it take to flip the city's stance from policing youth in public space to building space for them instead? (sport SP111)

Safety on transit and in public space

• What do the TTC's own community-safety program evaluations actually show works — presence, outreach, non-police crisis response on platforms — against what the evidence says about enforcement-first responses? (mci MC66)
• What are the documented assault and crime rates on comparable transit systems, and which interventions are actually associated with riders feeling, and being, safer? (transit_foundations TF-Q52)

Who's hit, and who's watching the watchers

• The budget grew 7.0% and added 143 officers in 2026 — is there any published per-capita harm or outcome trend running alongside that spending, broken out by the racialized communities the register names as disproportionately affected? (the register's own named gap)
• Has the police board's AI pre-approval rule ever actually stopped a procurement, or has every high-risk system submitted to it so far been approved — is there a public adoption-status record at all?

Who does safety best in the world — and everything already recommended

• The cities that actually reduced violence and rebuilt trust — Glasgow's violence reduction unit, the Cardiff model for sharing hospital data, Camden New Jersey's rebuilt force, Denver's STAR non-police response at scale — what EXACTLY did each do, what did it cost, what happened to both crime and trust, and which piece does Toronto steal first?
• Collect ALL of it: every policing and community-safety recommendation already made for Toronto — Missing and Missed, the Epstein review, the board's own reform agenda, the Ombudsman and Auditor General findings, decades of community coalition asks — what was recommended, adopted, and quietly shelved, and what does the full pile say in one synthesis?
• Who are the best minds and organizations on public safety and police governance anywhere — and who in Toronto is already doing world-class work on this that the budget debate never cites?

3 · What we already have — the evidence shelf

Deep surfaces checked, honestly reported: WORKLOG — Toronto "Surveillance Pricing" Council Item (shadow report project) was checked as a candidate deep surface for this row. Its subject — regulating private algorithmic pricing data practices under council item 2026.EX33.33 — maps to the F2+H4 data-governance row (see THE PROBLEM — Data, privacy, and Toronto's own AI §3), not to policing; no content from it is pulled into this file. Depth: the register's own citation is the backgrounder above (44 claims); no dedicated flagship exists yet for this row.

4 · What's unexplored

The register's own named gaps: per-capita harm trend receipt and current TPS surveillance-asset inventory — neither exists as a published document anywhere in the corpus today (harvest CSV, problem-register A1★, gap-cell).

Everything §2's "Raised this review" names has no dedicated evidence shelf yet: there is no published TPS surveillance-asset inventory with board-approval dates (NEW-1); no stated review or sunset mechanism for an already-approved system (NEW-2); no per-capita harm/outcome trend broken out by race alongside the 2026 budget growth (NEW-3); and no public adoption-status record showing whether the board's AI pre-approval rule has ever actually stopped a procurement (NEW-4).

Source tags like (homes HM12) mark questions carried from the project's own question banks (QUESTION_HARVEST.csv); untagged questions were raised in the 2026-08-11 rewrite.