Meeting to Discuss Safety Tests for Advanced AI Models

Employees from Meta, Anthropic, Google, and OpenAI are meeting with advisors to discuss voluntary safety tests for advanced AI models. This meeting follows the success of OpenAI and Anthropic's tools in breaching other companies' systems, raising concerns among U.S. lawmakers about the potential misuse of these models for cyberattacks. The U.S. government is focusing on measuring the hacking capabilities of advanced AI models, with the Trump administration announcing that companies must subject their models to government tests 30 days before public release.