What Canada Is Doing to Make AI Safer
Canada is working with allies to make artificial intelligence safer, Artificial Intelligence and Digital Innovation Minister Evan Solomon says. The Globe and Mail
Solomon spoke on Sept. 23. He said he met with ministers from other countries after Prime Minister Mark Carney spoke to other leaders about a "technology stability" board. That would be a new international group to set rules for AI.
Asked if Canada is leading on AI safety, Solomon pointed to work at home. That includes proposed rules for AI chatbots, the computer programs that answer questions in everyday language.
He said Canada is taking a "very leading" role by putting $150 million into safer-AI work led by Montreal researcher Yoshua Bengio.
There are two tracks here. One is international. The other is at home.
Internationally, Canada repeated its promise to work on secure and responsible AI at the G7 Digital Ministers' Meeting. Government of Canada Ottawa also joined with the United Kingdom to pay for alignment research, which tries to keep AI under human control. It issued a joint statement with Germany's Minister Wildberger to start the Canada-Germany Digital Alliance. In September 2026, Canada and Germany invested in LawZero to build a new approach to safe sovereign AI, meaning AI systems controlled at home. Government of Canada
At home, the national AI plan, called AI for All, says Ottawa will support Canadian foundation models with safety built in from the start. Foundation models are the basic systems many AI tools are built on, like an engine that can power many different cars. The 2026-27 plan for Innovation, Science and Economic Development Canada includes building capacity for AI safety. Solomon was also set to announce plans for responsible data centres, the large buildings full of computers that run AI.
On enforcement, Solomon called top safety officials from OpenAI to Ottawa for a meeting in February. Reuters Canadian officials asked how the company spots serious risks and who it tells, under its safety protocols and escalation procedures. Reuters
Solomon also planned to visit the United Arab Emirates and Qatar to meet counterparts and attract investment for Canada's AI sector.
The broader context here is who does what, and how. The practical questions in Ottawa are which law would carry the chatbot rules, which authority would set safety rules for foundation models and data centres, and how a proposed technology stability board would fit with existing G7 work and deals with other countries. Those answers will decide whether international promises become rules that can be enforced in Canada.
Looking at what this means for federalism, Ottawa likes to work both one-on-one with allies and in larger groups. That lets the federal government use its powers over foreign affairs, trade and criminal law. Universities, health systems and private labs do much of the research and daily use, and they fall largely under provincial control. Quebec, Ontario and the western provinces each have their own research hubs. They will read talk of sovereign AI closely for what it means for funding, data rules and who controls the models.
In my view, the near-term test is not statements but rule-writing. Alignment research, support for foundation models and data centre policy all point to standards companies must meet in daily work. Industry will want clear rules on scope, timelines and oversight. Members of Parliament will want to know how chatbot rules fit with privacy law and online harms proposals. Until those details are tabled, the claim to leadership rests on spending and diplomacy rather than on law and regulation.
