Mythos complicates the breakup, says Pentagon CTO, but Anthropic is still barred
Emil Michael says agencies are evaluating the cybersecurity model, not deploying it
Pentagon CTO Emil Michael pushed back on reports of a thaw in the department’s relationship with Anthropic: The two are not getting back together, even as Mythos draws interest from government agencies.
The DoD CTO told CNBC's Becky Quick on Friday that, as far as his department is concerned, Anthropic is still a supply chain risk that it doesn't want in its systems. He didn't appear to believe things were any different with regard to Anthropic elsewhere in the federal government, but confirmed recent reports that some agencies have accessed Mythos for evaluation purposes rather than operational deployment.
"The Mythos issue … is a separate national security moment," Michael said. "We have to make sure our networks are hardened up because that model has capabilities that are particular to finding cyber vulnerabilities and patching them."
Rumors that the administration's stance toward Anthropic, following an acceptable use dispute with the Pentagon, was softening have been swirling since mid-April, when it was reported that the National Security Agency was using Mythos. That, in turn, fueled speculation that the government might move toward broader adoption of the model, reinforced by CEO Dario Amodei's visit to the White House last month, but Michael insists that's not the case.
"With Anthropic, they're a supply chain risk," Michael said in the interview on Squawk Box.
"From a national security standpoint, you always have to look at those things. The NSA and Commerce evaluates all frontier models, including Chinese frontier models, to see what the capabilities are at the edge."
- Fewer users, fatter wallets is why Anthropic tops OpenAI in LLM revenue stakes
- Anthropic's super-scary bug hunting model Mythos is shaping up to be a nothingburger
- Anthropic admits it dumbed down Claude when trying to make it smarter
- Anthropic tests how devs react to yanking Claude Code from Pro plan
In other words, the official line from the administration at this point seems to be that any use of Mythos at a federal agency is just model analysis, and thus business as usual.
That doesn't mean the government isn't seeking to make use of any similarly capable models that might arise, like ChatGPT 5.5-Cyber, as AI leaders from multiple firms were reportedly planning to meet at the White House soon to discuss Mythos and the cybersecurity risks posed by emerging AI models.
"We think about the first drop being Mythos, but there's going to be others," Michael said on CNBC. "The government's looking for how to work with all of these companies in the coming year so that their capabilities are understood by us first so that we can fix any issues we have in the private and the public sectors."
Neither the Pentagon nor Anthropic responded to questions for this story. ®
