Canada signs the Paris Call for AI safety, alongside OpenAI, Anthropic and Cohere

Canada signs the Paris Call for AI safety, alongside OpenAI, Anthropic and Cohere

Canada has signed the Paris Call for AI Safety and Security, a voluntary declaration announced on Nov. 9. Six governments signed: Canada, France, Finland, Norway, the Netherlands and Bahrain. Ten AI companies joined them, including OpenAI, Anthropic, Microsoft, Mistral AI, Hugging Face and Canada’s Cohere.

What signatories promise

  • Testing and risk management: a common framework to evaluate frontier models, with a focus on misuse such as cyberattacks and on loss of control, and published, non-sensitive evaluation results.
  • Accountability: developers disclose capabilities, limits and known risks, support independent evaluation and keep shutdown mechanisms for models they control.
  • Early warning: rapid alerts and information sharing on serious incidents, plus secure channels for whistleblowers.

The other side

The biggest AI powers are missing. The U.S. and Chinese governments did not sign, and neither did Google or Meta. The declaration is voluntary, with no enforcement mechanism, so its impact depends on whether more countries and companies join. It remains open to new endorsements.

Why Canada should care

Cohere is the only Canadian company on the list, and Ottawa is investing up to $1.6 billion in it. The commitments could shape how AI tools used by Canadian businesses are tested and how incidents are reported.

For businesses buying AI tools, the declaration offers a useful checklist: ask vendors how they test their models, how they report incidents and how they would shut a system down if it misbehaved.

Sources: Prime Minister’s Office, BetaKit, Government of the Netherlands, University of Toronto.

Leave a Reply

Your email address will not be published. Required fields are marked *

Most Read

Subscribe To Our Magazine

Download Our Magazine