The White House Office of Management and Budget is preparing to authorize a modified version of Anthropic's Claude Mythos model for use by US federal civilian agencies. The model, announced under Project Glasswing, can identify thousands of zero-day vulnerabilities across major operating systems and browsers. The OMB is working with Anthropic and the intelligence community to establish guardrails including air-gapped environments and human-in-the-loop review. This move creates a civilian workaround to the Pentagon's existing supply-chain risk designation that bars Anthropic from defense contracts. Experts note the situation highlights the need for multi-layered control frameworks for potent AI capabilities and sets a precedent for broader government and enterprise adoption.
Sort: