Protesters broke Mila's windows on the eve of Canada's biggest AI summit
The AI backlash stopped being a polling number this week. In Montreal it smashed about 20 windows, and by Wednesday evening police were pulling people out of a march outside the country's largest AI conference. Hours later and 4,000 km away, California put a kill switch on the agenda.
About 10 people smashed roughly 20 windows and spray-painted graffiti at the Montreal offices of Mila, the AI institute founded by Yoshua Bengio, late Tuesday night — hours before the city opened ALL IN, billed as Canada's largest AI and technology conference. Police arrested a 25-year-old woman near the scene and are reviewing surveillance footage. CBC reports a second building, a software company, was tagged the same night. Then on Wednesday, an anti-AI march organized by the group Résistance Montréal along Jeanne-Mance Street, a block from the Palais des Congrès where the conference was running, turned violent; police made one arrest, a 19-year-old man.
Bengio addressed it from the stage. "Violence is not an answer to these risks and fears, which I can understand," he said. Mila condemned "any use of violence, property damage and tactics aimed at intimidating our staff and researchers." Federal AI Minister Evan Solomon, at the conference, said protest is welcome in a democracy but that the safety and security of researchers, students and workers would be protected.
That is the trade the backlash just made for itself. Public anger at AI is real and rising — jobs, water, data centers, and an Anthropic safety researcher's resignation warning the technology could kill humans all put the case on front pages this month. Wrecking an institute's windows and brawling outside a conference hall converts a political argument into a police file, and hands the industry a victim story at the exact moment regulators in Sacramento and Washington are deciding how hard to push. The hardest questions about AI are about corporate power and unaccountable labs; broken glass answers a different question.
Also Friday: California Governor Gavin Newsom signed an executive order directing the state to "dramatically accelerate" independent, third-party oversight of AI companies and to advance the creation of an "AI kill switch" for frontier models. A panel of experts has two months to deliver a guide for strengthening the state's AI safety laws, with options including third-party-written safety plans for frontier labs and mandating a shutdown mechanism for emergency use. We covered the build-up on Thursday — Newsom floats special session or executive action on AI safety — and Friday he picked the fastest door: no floor vote, no deal with the industry lobby, no help from a Congress that has gone home to campaign.
Read the order precisely, though: it does not yet impose a kill switch on anyone. It orders a study with a deadline, and the state's binding AI law — the third-party auditor framework Newsom signed on September 9 — has none. The gap between a two-month guide and a rule labs must obey is where this fight lives, and the kill-switch idea is technically underspecified: shutting down a frontier model mid-training is not a switch on a rack, it is weights, clusters and a rollout. Whose hand is on it, and what counts as an emergency, is the whole question — and the order punts it to a panel. Newsom, whose term ends January 4, is openly mulling a 2028 run and now has a signed document to point at.
What to watch: whether the two-month California guide lands as an enforceable mandate or another set of recommendations — and whether the labs that keep asking for regulation fight the kill-switch clause once it is written down.
Is street-level protest helping the AI safety cause or handing the labs a victim story? Tell us in the comments.
