The White House is in active negotiations with Anthropic to deploy the AI company's new Mythos Preview model across federal agencies, even as the Pentagon wages an ongoing legal battle to blacklist the firm as a supply chain risk.
The discussions reveal a fundamental split within the Trump administration. While some officials have embraced the Pentagon's designation of Anthropic as a security threat, others across civilian agencies view the company's technology as essential to protecting critical U.S. infrastructure.
Anthropic is rolling out Mythos to a carefully selected group rather than releasing it publicly, allowing organizations to evaluate its powerful cyber capabilities and shore up defenses. Government agencies have begun requesting access, prompting the White House and Anthropic to work through the terms under which deployment might occur. The Office of Management and Budget signaled it is reviewing the matter in response to agency inquiries.
Multiple sources indicate agencies could gain access within weeks, though progress remains uneven. "There's progress with the White House. There's not progress with the Department of War," one administration official said.
The Pentagon stalemate
Anthropic remains locked in litigation with the Pentagon, which declared the company a supply chain risk and ordered military contractors to strip its software from defense workflows. The company is barred from Pentagon contracts but retains the ability to work with other government entities during the legal proceedings.
The Pentagon's core objection centers on Anthropic's stated refusal to allow its models for mass surveillance or fully autonomous weapons development. Defense officials say these restrictions are too vague and demand assurances that the military can deploy AI systems for all lawful purposes.
One Defense Department official acknowledged the underlying reality during the height of tensions: the only reason talks continued at all was "these guys are that good."
Departments of Energy and Treasury are particularly eager to access Mythos. These civilian agencies oversee critical infrastructure including the electrical grid and financial system. The model could help identify vulnerabilities to cyberattacks and prepare defenses before attackers strike. Such work falls outside the military applications that triggered Pentagon objections.
Some administration officials bristle at what they view as Anthropic's scaremongering. One government official accused the company of weaponizing concerns about Mythos's hacking capabilities to cultivate support within the bureaucracy. "They're using this Mythos cyber weapon to find friendly ears in the government," the official said. "They're succeeding."
Yet other officials defend the need for unrestricted evaluation. "The government has a responsibility to evaluate every model to see where the frontier of tech is," another administration official argued. Intelligence agencies already rely on Anthropic's existing models, and civilian agencies recognize the company's tools as among the most capable available for national security applications.
The dispute reflects a broader tension within the administration. Some Trump officials view Anthropic's leadership as ideologically hostile, viewing the company's ethical guardrails as part of a "woke" agenda. These officials relished applying the supply chain risk label. But many of those same skeptics, and numerous others, cannot ignore that Anthropic's technology represents the current frontier of AI capability for security purposes.
Author James Rodriguez: "The Pentagon's refusing to budge on access while civilian agencies circle the plate shows how badly Washington compartmentalizes security risk, and how little some military brass can stomach being told no by a tech firm."
Comments