AI company leaders urge UN coordination on safety standards
Technology & SocietyWell-covered
|
The Facts
- Sam Altman, Dario Amodei and Clément Delangue addressed the UN Security Council on AI.
- Altman called for international cooperation on AI governance.
- Altman called for common benchmarks to measure AI capabilities and safety measures.
- Amodei proposed shared standards for testing AI models for loss-of-control and misuse risks.
- The executives said AI risks require governments to act together across national borders.
- U.S. representatives opposed a new global AI governance structure at the UN meeting.
- Governments remain divided over whether AI should face global limits or primarily market-led development.
Context
What standards did the AI leaders seek?
Altman advocated shared benchmarks for assessing AI capabilities and company safety measures. Amodei called for standards to test models for loss-of-control and misuse risks, alongside systems for verification and incident reporting. Московский … BBC uol.com.br
Why did they ask for international coordination?
The executives said increasingly capable AI could advance faster than governments and institutions can respond, and that major decisions should involve accountable governments rather than a small number of companies. Washington Post uol.com.br MoneyControl
What remains unresolved?
Countries have not agreed on whether to establish global AI controls. U.S. officials opposed a new global governance structure, while AI company leaders urged international safeguards. India Today BBC CNBC
Where Left and Right agree, and where they split
- Where Left and Right agree
- Both accept that Altman and Amodei proposed concrete benchmarks and shared testing standards for loss-of-control and misuse risks, and that U.S. representatives rejected a new global structure.
- Where Left and Right split
- Governments failing to convert safety proposals into binding, enforceable rules, or correctly keeping AI oversight under national control and market discipline.
- Why they won’t converge
- The divide is about trust in institutions: whether an enforceable body can bind AI development to democratic accountability, or whether any global structure would dilute national control without proving it works better.
How left and right read it
Only governments can write binding rules. That is why Altman's call for common benchmarks and Amodei's for shared testing against loss-of-control and misuse risks matters — and why the U.S. representatives who rejected a new global structure owe a reason for leaving safety to the market. Security Council speeches are cheap; what counts is enforceable standards, negotiated publicly, not lobbying that points the other way.
“The OpenAI CEO's speech contrasted with some of his company's lobbying efforts and the political action committee of its president, Greg Brockman.” — The Guardian
Authority over American technology should stay in American hands unless someone can prove an international body would do better, and no such proof was offered at the Security Council. Altman asked for international cooperation and common benchmarks while Amodei urged shared testing standards for loss-of-control and misuse risks, yet firms are free to adopt such benchmarks themselves without a new global structure. U.S. representatives were right to say no. Keep the presumption with national control and market-led development.
The receipts — all 100 sources
Wire services (3)
Independent coverage (50)
Facts first. Then every angle.
The day’s biggest stories in one short brief — the facts everyone agrees on, then the competing values behind the headlines. Free in your inbox.