The White House is striving to integrate Mythos, the powerful and controversial new artificial intelligence (AI) model, into the federal government’s arsenal even as the Trump administration remains locked in a venomous legal and political battle with the model’s creator, Anthropic.
In a memo reviewed by Bloomberg News, Gregory Barbaccia, the federal chief information officer at the Office of Management and Budget (OMB), informed Cabinet officials on Tuesday that the administration is establishing safeguards to allow agencies to begin using a modified version of Mythos.
While Barbaccia’s memo did not provide a specific timeline or a definitive list of agencies that will receive access, it told technology chiefs to expect further details “in the coming weeks.” For now, the U.S. government appears determined to harness the power of Mythos, even if it means doing business with a company it has labeled a threat.
The email, sent to the Departments of Defense, Treasury, State, and Justice, could indicate that the government is working with the intelligence community to ensure “appropriate guardrails” are in place before a wider rollout.
The move signals a pragmatic, if awkward, pivot for the Trump administration. Anthropic is currently suing the government after the Pentagon designated the AI lab a “supply chain threat,” a label typically reserved for foreign adversaries. The blacklisting followed a contract dispute in which Anthropic reportedly refused to allow its Claude AI model to be used for “all lawful purposes,” including autonomous weaponry and mass surveillance.
The tension reached the highest levels of the West Wing on Friday, when Anthropic CEO Dario Amodei met with White House Chief of Staff Susie Wiles. Reports suggest the administration is attempting to balance its hardline stance against Anthropic’s corporate policies with the undeniable national security implications of the Mythos technology.
“The federal government’s push to adopt Mythos reveals that elite AI capabilities have transitioned from optional upgrades to essential pillars of national security, transcending the executive branch’s personal and legal agendas,” HyperFRAME Research analyst Ron Westfall said in an email. “By developing a restricted security enclave for a sanitized version of the model, the OMB is carving out a functional workaround to the Pentagon’s strict embargo, exposing a fracture between political leadership and the operational needs of civilian cyber-defense.”
“From my viewpoint, the move underscores a cold reality in the 2026 geopolitical landscape: when a technology offers a decisive tactical edge in the AI arms race, strategic survival will consistently take precedence over even the most heated domestic political warfare,” Westfall added.
Mythos has sparked equal parts wonder and alarm within the intelligence community. Developed under Anthropic’s Project Glasswing, Mythos has demonstrated an unprecedented ability to identify thousands of major vulnerabilities in operating systems and web browsers.
While officials view the tool as a revolutionary asset for defensive cybersecurity, experts warn that its advanced coding capabilities could be weaponized by hackers to “map out” catastrophic attacks against government infrastructure.
Understandably, Anthropic has been extremely protective of Mythos, limiting its release to a select group of financial firms and tech companies. Within the company, leadership reportedly feared the model posed a national security risk after internal testers used it to uncover critical bugs that usually require the world’s most elite human hackers to find.
Despite the ongoing litigation, including a recent court order won by Anthropic to block a government ban on its technology, the Treasury Department and the Federal Reserve have already urged Wall Street leaders to use Mythos to shore up their defenses.
