Eight countries and top AI labs agree to shutdown mechanisms in new Paris declaration
The Decoder
Eight countries and leading AI labs signed the Paris declaration agreeing to safety measures like shutdown mechanisms.
Representatives from eight countries, civil-society organizations, and major tech companies adopted the Paris Call for AI Safety and Security ahead of the Paris Peace Forum. The political self-commitment includes shared evaluation procedures, developer accountability, and the implementation of shutdown mechanisms for powerful AI models.
The agreement involves notable industry figures such as Sam Altman of OpenAI, Dario Amodei of Anthropic, Brad Smith of Microsoft, and Arthur Mensch of Mistral AI, alongside French President Emmanuel Macron and U.S. Treasury Secretary Scott Bessent. Germany did not sign the declaration.
While the document outlines concrete safety steps, it does not currently establish a binding international oversight system, deadlines, or sanctions. The initiative responds to growing concerns regarding model controllability, cyberattacks, and recent incidents involving autonomous AI actions.
Despite tensions and differing U.S. approaches to international oversight, France also plans to advance a child-protection charter at the forum. This framework aims to fund independent research into AI impacts on youth cognitive development and emotional attachment.
Why it matters: The declaration represents a foundational political step toward international AI safety standards and accountability among major labs, despite lacking binding enforcement mechanisms.



