New to this? Read it in simple words
- Countries and ten AI companies signed a safety pledge in Paris on November 9.
- The companies promise ways to shut down their own AI models if needed.
- The pledge is voluntary and sets no deadlines.
- Google, Meta, the US and China did not sign.
- Fail-safe
- A way to stop a system safely when something goes wrong.
- Whistleblower
- A worker who reports wrongdoing or danger inside their own organisation.
- Voluntary
- Done by choice, with no law to force it and no penalty for breaking it.
Countries have signed many statements on AI safety since 2023. This one stands out because AI companies signed the same text as governments. About 50 people joined Macron’s talks, many by video, including OpenAI’s Sam Altman and Anthropic’s Dario Amodei, TNW reports.
Who signed
Six countries signed on November 9: Bahrain, Canada, Finland, France, the Netherlands and Norway, Canada’s government says. The French presidency’s page now lists 16 countries, and says the text stays open to new signers.
Ten companies signed: Anthropic, Black Forest Labs, Cohere, Domyn, Hugging Face, Microsoft, Mistral AI, OpenAI, Sakana AI and Synthesia. Seven research and civil-society groups also signed, among them LawZero and SaferAI.
From the text and the signer lists. Six countries signed on the day.
The shutdown promise
“Developers also commit to implement fail-safe mechanisms to shut down model operations within their control, including during training and internal use,” the text says. It gives restrictions, pauses or shutdowns as examples.
The promise covers only systems the companies control themselves. It is not a switch that governments could use.
What else it says
Signers will build a common way to test and manage the risks of the most advanced models. They aim to publish non-sensitive test results in a standard format, in a public shared library.
Developers will disclose what their models can do, their limits and known risks. Governments promise safe channels for whistleblowers, and clearer rules on who is liable for large-scale harm.
The text also calls for quick alerts on serious cyber and AI control incidents. Signers will only explore whether a joint system for this would help.
What it lacks
The pledge is voluntary and sets no deadlines. “It does not spell out whether or how that would be enforced,” The Globe and Mail reported.
CeSIA, a French AI safety group that signed, wants public authorities to have the power to suspend a model, TNW reports. Google, Meta and SpaceX did not sign.
The US government did not sign, though Treasury Secretary Scott Bessent took part in the talks, the German news agency DPA reports. China did not take part.
Sources
Every fact in this story comes from the sources below. Open them to check our work.
- 1
- 2Primary source · November 9, 2026Joint declaration on Paris call for AI safety and security Prime Minister of Canada
- 3Research · November 10, 2026AI firms sign Macron’s Paris Call, but safety group says it lacks teeth TNW
- 4Research · November 10, 2026Canada joins OpenAI, Anthropic in AI standards pledge The Globe and Mail
- 5Research · November 10, 2026Government of Canada signs international declaration on AI safety and security BetaKit
- 6Research · November 9, 2026Macron pushes for tighter AI safety rules at international talks DPA via Yahoo News
The commitments are quoted from the text itself. The list of countries is still growing: six signed on the day, and the French presidency’s page now names 16.
AI labs promise shutdown plans in a new Paris pledge. Silicon AI News, November 10, 2026. https://siliconainews.com/stories/paris-call-ai-safety-security
Ten AI labs signed a written promise to build shutdown mechanisms, but the pledge is voluntary, unenforced and lacks Google, Meta, the US and China. How we rate the mood
The mood shows how the news frames the event for the tech and AI world: as an opportunity (6 and above) or as a loss (below 5). It is not our verdict on who is right.