U.S. Expands Government Stress Tests to Cover Google DeepMind, xAI and Microsoft AI Models
The U.S. government has broadened a program that gives federal AI researchers access to unreleased models for risk assessment, adding Google DeepMind, xAI and Microsoft to participants. OpenAI and Anthropic were already collaborating with the U.S. Center for AI Standards and Innovation to test models for vulnerabilities. The scientists are focused …