Overview
- The Pentagon named seven companies Friday to provide AI models for its highest classified operations, excluding Anthropic.
- The Defense Department says the tools will speed data fusion and context understanding for troops, while humans keep control over strike decisions.
- White House officials oppose Anthropic’s plan to expand access to its new Mythos model to about 120 users, citing security risks and limited computing capacity reported by the Wall Street Journal.
- Anthropic says it is investigating an unauthorized access to Mythos through a third-party contractor and has found no sign the breach went beyond that vendor so far.
- The government cut Anthropic contracts in late February over the company’s limits on mass surveillance and lethal uses, and the startup is challenging the decision in court.