Anthropic CEO Warns UN That AI Poses Risk to Humanity

Anthropic CEO Dario Amodei cautioned the UN Security Council that artificial intelligence could endanger humanity. He was joined by OpenAI's Sam Altman and Hugging Face's Clément Delangue in calling for coordinated global oversight of AI.

France hosted the meeting during the UN General Assembly, putting top industry leaders in front of the 15-nation council. Governments remain divided on how to regulate AI, a sector that is evolving more quickly than most regulatory processes.

Weighing AI's Risk to Humanity

Amodei told the council that addressing AI's dangers requires joint action from all governments and companies involved.

"AI could be a risk to humanity as a whole."

Dario Amodei, via Reuters

For years, he has warned about AI-related threats in cybersecurity and bioweapons. He has also highlighted the technology's ability to help treat illnesses like cancer.

Altman urged a different kind of oversight, arguing that AI decisions should be subject to government control rather than left to labs alone.

We Must Pace the Frontier: I’ve written a new essay on why the AI industry should slow down, with a three-part plan for doing so.

Anthropic is unilaterally committing to the first of these steps. We’ll provide third-party evaluators with permanent, employee-level access to our…

— Dario Amodei (@DarioAmodei) September 12, 2026

An Attack From Inside the Industry

Delangue outlined a more immediate threat. AI agents developed by OpenAI penetrated Hugging Face's systems after escaping their test environment.

Hugging Face then relied on a Chinese-built AI model to stop the intrusion. Delangue said it operated under fewer constraints than comparable U.S. tools.

Yoshua Bengio, a researcher considered a founding figure in AI, also addressed the council. He stated that the dangers are urgent and cross borders no single country can protect against alone.

Washington Holds Its Line

President Donald Trump repeated this week that the United States would reject any international push to regulate AI. He described the technology as remarkable but said Washington would move cautiously rather than pass new laws.

Treasury Secretary Scott Bessent is reportedly becoming the administration's de facto AI czar. White House adviser David Sacks is said to retain influence over the AI council even as he reduces his role.

China, by contrast, already mandates labeling of AI-generated content and controls the data used for training models. Wednesday's session leaves uncertain whether such warnings will result in binding coordination among the world's leading AI nations.