Anthropic is actively investigating an unauthorized access claim involving its Mythos AI model, which has been assessed as capable of identifying large numbers of zero-day vulnerabilities. A separate allegation that the NSA may be using Mythos outside a formal procurement agreement raises access governance questions for any organization with API integrations to Anthropic services. The access method is unconfirmed; no technical IOCs are available. This item warrants AI API access governance review rather than an immediate incident response action.