Overview
- Leaders of Anthropic and OpenAI warned the U.N. Security Council that the fastest AI models could grow so powerful they are hard or impossible to control.
- They urged multilateral safeguards, independent verification of model capabilities, and limits on concentrating AI power in a single company or country.
- Argentina’s president said he will not impose preventive AI rules and offered fiscal and legal incentives to attract AI firms, a stance that contrasts with the companies’ calls.
- A U.S. administration representative rejected a global regulatory scheme and recommended sharing best practices and building national capabilities instead.
- The debate raised security concerns about autonomous weapons, noted Anthropic’s pledge to slow development when needed, and set up competing agendas for the G20 and other diplomatic forums.