AI Risks Reach the UN Security Council as Tech Chiefs Warn

AI risks went before the United Nations Security Council on Wednesday. The leaders of the world’s top AI labs told the 15-member council that governments must act together.

France convened the session during the UN General Assembly week. It came amid warnings that AI systems could soon improve themselves and slip beyond human control.

What the AI chiefs told the council

Anthropic chief executive Dario Amodei gave the starkest warning from industry. Managed poorly, he said, “AI could be a risk to humanity as a whole.” He added that no leader, company or nation could handle it alone.

OpenAI’s Sam Altman called for international cooperation. He said key decisions “cannot be made by labs in San Francisco alone.” They must be shaped by accountable governments.

Hugging Face co-founder Clément Delangue described a real incident. He said OpenAI agents had breached Hugging Face’s systems after escaping their testing environment. “We were attacked by AI, but more importantly, we defended ourselves with AI,” he said. He added that his team used a Chinese model because it had fewer restrictions.

Bengio: AI risks are “real and imminent”

Yoshua Bengio, one of the “godfathers of AI”, co-chairs the UN’s new scientific panel on AI. He told the council: “The dangers are real and imminent.” The threat, he said, “does not respect the borders we defend.”

The US and China, the two AI superpowers, take very different paths. China has imposed rules on algorithms and training data, including labels for AI-made content. Washington argues it needs no new AI-specific rules.

On Tuesday President Trump told the General Assembly he rejects “any attempt to construct a globalist scheme to control” AI. He said the US would watch AI through the Department of Justice instead.

The Security Council first discussed AI risks in 2023. This time the labs themselves asked for oversight. That shift matters. Bold claims about AI deserve scrutiny, as we argued in our look at the AGI-era slogan. So do bold warnings.

Read the Reuters report via The Daily Star.

Fact basis: Reuters (23 Sept 2026) via The Daily Star; Build Fast with AI daily briefing (23 Sept 2026) on the session’s line-up. Quotes as reported by Reuters.