UN Security Council Will Get Advice on AI Risks From Tech Giants Building It

The U.N. Security Council will host briefings from several prominent AI companies and researchers this Wednesday as part of a session devoted to risks posed by advanced artificial intelligence. Representatives scheduled to speak include Sam Altman of OpenAI, Dario Amodei of Anthropic, Yoshua Bengio, and Clément Delangue of Hugging Face, alongside Chinese firms DeepSeek and Moonshot. France, which holds the council's rotating presidency this month, organized the meeting and will have Foreign Minister Jean-Noël Barrot chair it. The concept note for the session centers on scenarios in which powerful AI systems could operate beyond human control, touching on dangers such as AI-enabled cyberattacks, capabilities to sway elections, and automated weapon design without human oversight. Anthropic's CEO Dario Amodei will brief the council remotely. In a recent essay, Amodei urged a deliberate slowdown in developing frontier AI models and proposed giving outside evaluators ongoing, staff-level access to verify safety claims. He also described China as the autocratic country with the most advanced AI capabilities and urged continuation of U.S. export restrictions on advanced chips to Beijing—positions that drew criticism from Chinese officials and state media. The meeting is unlikely to yield binding agreements, mirroring other recent international gatherings that produced discussion but no public commitments. Its scheduling shortly before a high-profile U.S.-China summit in Washington underscores the session's geopolitical context as officials weigh both cooperation and competition in AI governance.