@thehill: OpenAI, Anthropic reach AI safety, research agreement with feds trib.al/1sFjjC0