NewsPolicy
Canada joins the Paris call for AI safety and security, alongside Anthropic, OpenAI, Mistral and Cohere
BetaKit reports that Canada announced on November 9, 2026 that it backs a joint declaration on AI safety and security known as the Paris call. Bahrain, Finland, France, Norway and the Netherlands have signed too, along with ten AI companies; BetaKit names Anthropic, OpenAI, Mistral and Cohere. Signers commit to allow some independent testing of their models, give early warnings, shut systems down when problems arise to prevent cascading risks, disclose the capabilities, limits and risks of frontier models, develop common technical standards, and monitor and report to limit harm. The US and Chinese governments, Google and Meta have not signed, and the declaration stays open to more endorsements.
Why it matters
It is a voluntary pledge, not a law, but it names the practices governments now expect from frontier labs: outside testing, early warnings and a way to shut a system down.