Here's an ironic twist: Anthropic's new Mythos model is designed to find security vulnerabilities, but the US agency in charge of coordinating cybersecurity can't use it yet.
According to Axios, the Cybersecurity and Infrastructure Security Agency (CISA) doesn't have access to Mythos Preview. Meanwhile, other federal agencies like the Commerce Department and National Security Agency are already testing it out.
Mythos is Anthropic's specialized model for security work, built to identify and help patch vulnerabilities in code. It's part of a broader push to use AI for defensive cybersecurity, not just offensive capabilities.
The Trump administration has reportedly been negotiating broader access across agencies. But the fact that CISA, the central hub for national cyber defense, is sitting this one out feels like a coordination miss.
For anyone using AI tools in security work, this highlights how fragmented adoption can be, even within government. The technology might be ready, but access and deployment are still political and bureaucratic processes.
It's worth watching how this plays out. If federal agencies can't coordinate on AI security tools internally, it raises questions about how they'll help private sector organizations do the same.
