AI Generated Image
U.S. Presses Meta to Agree to A.I. Reviews
The New York Times • • business
Key Points:
- The Trump administration is urging Meta to voluntarily submit its artificial intelligence models for government review to assess their capabilities and vulnerabilities, marking increased oversight efforts in the AI sector.
- Meta remains the only major U.S. AI developer yet to agree to share its models with the federal government, while companies like OpenAI, Anthropic, Google, xAI, and Microsoft have complied.
- The government’s AI safety group, the Center for A.I. Standards and Innovation (CAISI), is actively engaging with companies to establish voluntary agreements for model review.
- Meta has expressed willingness to cooperate and hopes to finalize an agreement soon, emphasizing shared goals of advancing secure and robust AI leadership in the U.S.
- Growing concerns about cybersecurity risks related to advanced AI models have prompted government action, although some industry experts argue that AI can be used defensively as well as offensively.
