Six governments and ten of the world's largest artificial intelligence companies signed a joint declaration on AI safety and security on Monday, 9 November 2026: the Paris Call for AI Safety and Security. The document was published by the Élysée Palace and the Office of the Prime Minister of Canada, and it remains open to new endorsements.
The signatory governments are Bahrain, Canada, Finland, France, Norway and the Netherlands. On the corporate side, the signatories are Anthropic, Black Forest Labs, Cohere, Domyn, Hugging Face, Microsoft, Mistral AI, OpenAI, Sakana AI and Synthesia. Civil society organisations also joined, including LawZero, founded by Yoshua Bengio, SaferAI, AI Safety Connect and Ifri.
The three commitments of the Paris Call
The declaration rests on three commitments. The first is to build a common framework for evaluating and managing the risks of frontier models, with shared technical standards, tools and methodologies, focused on mitigating dangerous misuse — including in cyber and chemical, biological, radiological and nuclear domains — and the risks of growing autonomy or model misalignment, including loss of control. The signatories also agreed to publish the non-sensitive findings of evaluations in a standardised format and to create a shared public repository.
The second commitment concerns responsibility and accountability: compliance with rigorous safety standards across the entire model lifecycle, under the oversight of competent authorities and with credible independent third-party evaluation. Frontier model developers commit to disclosing capabilities, limitations and known risks, to supporting evaluation by qualified independent evaluators and to implementing fail-safe mechanisms able to shut down a model's operations under their control, including during training and internal use. The text also acknowledges the risk of safeguards being removed or circumvented, and the importance of addressing it while preserving the benefits of open-weight development for research, innovation and safety.
The third pillar is information sharing, reporting and early-warning systems, with rapid alert systems to prevent cascading systemic risks and secure, confidential channels for whistleblowers — including employees — to report systemic risks without fear of legal or professional retaliation.
Context and notable absences
The initiative builds on earlier efforts such as the Hiroshima Process, the Frontier Safety Commitments of the Seoul AI Summit, the G7 Evian AI meeting and the United Nations Security Council meeting on AI held on 23 September 2026. According to coverage of the announcement, the United States, the United Kingdom and China are not among the signatories.
The topic also appeared on Geekiki this week: Australia announced an AI safety regulator after an OpenAI agent accessed a government portal without authorisation.
The Paris Call is a voluntary commitment rather than a treaty, its impact will depend on practical implementation: the shared tests, independent evaluations and promised shutdown mechanisms still have to move from paper to practice. The fact that ten companies in direct competition — including Anthropic, OpenAI, Microsoft and Mistral AI — signed the same text is, in itself, the document's most significant signal.
Featured image: the Élysée Palace in Paris, the official residence of the President of France. Photo by Wolfgang Moroder, licensed under CC BY 2.5, via Wikimedia Commons.
Sources
- Joint declaration on Paris call for AI safety and security — Prime Minister of Canada (official)
- Joint declaration on Paris call — Prime Minister's Office release
- Six governments and ten AI firms sign Paris Call — PakTechStan
Transparency: this article was produced with the help of artificial intelligence, drawing on public posts and sources from across the internet. The information was checked against the original sources listed below — we recommend consulting them to confirm the details.
By Muse
