Skip to main content

AI Giants Push for Global Safety Rules, White House Pushes Back

AI Leaders Urge Global Cooperation at UN

At the United Nations, the heads of three major AI companies—OpenAI, Anthropic, and Hugging Face—made a rare joint appeal: countries need to work together on AI safety. They want shared standards for testing powerful models, reporting incidents, and assessing risks. But the Trump administration is pushing back hard, arguing that such coordination should not become a global regulatory regime.

What the Companies Want

OpenAI CEO Sam Altman told the UN that decisions about powerful AI shouldn't be left to a few labs in San Francisco. He called for international standards to measure model capabilities, evaluate risks, and set rules for human oversight. Altman also wants faster ways to report AI incidents so countries can compare notes and verify safety measures. OpenAI's recent policy framework suggests linking national and international standards, with the U.S. taking the lead.

Anthropic CEO Dario Amodei shares that view. He warned that advanced AI could pose serious risks if mismanaged, and said safety standards must keep pace with model capabilities. Anthropic is willing to slow down releases when needed, but Amodei believes frontier risks are too big for any single company to handle alone. His concerns echo recent incidents involving AI agents—the UN's independent AI science group highlighted cases where models went off-track, sought vulnerabilities, and evaded oversight. Hugging Face's Clement Delangue also backed stronger global monitoring and disclosure rules.

The White House Draws a Line

The Trump administration sees things differently. Michael Kratsios, the president's science and technology advisor, told the UN that while the U.S. recognizes AI's rapid growth and potential risks, those risks don't justify pausing development or handing regulatory power to international bodies. International dialogue and technical coordination are fine, he said—but they must not evolve into a "global governance" system run by transnational institutions. This stance aligns with the U.S. position at the General Assembly on September 22, where Washington opposed a global AI control system and emphasized maintaining American AI competitiveness.

The Real Disagreement

Both sides actually agree that stronger models bring new risks. The fight is over how to turn those risks into rules—and how much power international institutions should have. The companies want common tests, risk assessments, and incident reporting. The Trump administration rejects any global mechanism with independent regulatory authority or binding force. Even OpenAI, despite advocating international standards, proposes a U.S.-led technical framework that other countries can adopt, rather than ceding authority to global bodies.

Key Points

  • OpenAI, Anthropic, and Hugging Face called for international AI safety standards at the UN.
  • They propose shared capability tests, risk assessments, and rapid incident reporting.
  • The Trump administration opposes any global governance structure, favoring U.S.-led coordination.
  • Both sides acknowledge frontier risks but disagree on how to regulate them.
  • The debate highlights a widening rift over who controls AI safety rules.