BitGo CEO Mike Belshe has rejected a viral claim that Anthropic’s Mythos model breached nearly all of the National Security Agency’s classified systems, calling the story false as it spread across X this weekend.
His pushback targets posts that recast the government shutdown of a three-day-old model as a real-world hack. The fuller record is less dramatic.
Where the Mythos NSA Breach Claim Came From
The claim originated with Senator Mark Warner, vice chair of the Senate Intelligence Committee. The Economist reported his account of what the NSA director told him.
Warner said General Joshua Rudd, who leads the NSA and US Cyber Command, described the tool in stark terms.
“This tool broke into almost all of our classified systems, not in weeks, but in hours,” the Economist wrote, citing Warner.
Warner raised the example while praising Anthropic, not condemning it. He used it to argue for faster pre-release testing of frontier models.
The detail that went missing online is simple. This was an authorized red-team test on the agency’s own networks, not an outside intrusion.
Shashank Joshi, the Economist editor who published the quote, later cautioned it should not be read literally. He said it depended on Mythos working alongside other tools in particular conditions.
The US government was already a Mythos partner. Anthropic had deployed the model to government cyber defenders through Project Glasswing since April.
Belshe and Others Question the Framing
Belshe, the co-founder and chief executive of digital-asset custodian BitGo, answered one of the threads bluntly.
“I’m calling BS on this,” he challenged.
Follow us on X to get the latest news as it happens
He was not alone. Zack Korman mocked how the claim moved from senator to journalist to social media unchecked.
Analyst Kyle Chase noted the break-in was a test. He said a separate jailbreak flagged by Amazon was the real trigger.
Anthropic’s own statement supports them. It said the flagged jailbreak simply asked the model to read a codebase and fix flaws.
The technique surfaced a few minor, already-known bugs that rival models like OpenAI’s GPT-5.5 can also find.
The company disabled both models on June 12 to meet a US export-control directive, not because of any battlefield breach. It objected to recalling a model used by hundreds of millions of people over one narrow flaw.
Whether the test justified pulling the models is still contested. AI researcher Pedro Domingos argued the export controls were responsible, given the model’s powerful hacking capabilities.
Anthropic itself calls Mythos the strongest cyber model in the world. Yet it says recalling a tool over one flaw would freeze new releases across the industry.
The company is now working to restore access, and is drafting a shared risk framework with the White House.