Pentagon Rules Out Reconciliation with Anthropic After Mythos Dispute, Vows Multi-Model AI Strategy

The Pentagon's top technology leader said Thursday that the Department of Defense will not restore its working relationship with Anthropic, even as the White House has taken a notably softer tone toward the AI company in recent days.

Emil Michael, under secretary of Defense for research and engineering, made the remarks during a fireside chat at the AI+ Expo, where he described the fallout from the Mythos episode as a broader inflection point for how the military procures and relies on AI systems.

"Never again will we be single-threaded with any one model," Michael said, signaling that the Pentagon intends to diversify its AI dependencies rather than anchor critical capabilities to any single provider.

Michael characterized the dispute as something of a "cyber moment" — a phrase invoking the kind of systemic reckoning that has historically prompted lasting structural changes in how institutions think about digital risk and vendor concentration.

The remarks came the same day White House chief of staff Susie Wiles addressed AI policy more broadly, telling reporters that the Trump administration is "not in the business of picking winners and losers" on AI.

"This administration has one goal; ensure the best and safest tech is deployed rapidly to defeat any and all threats," Wiles said in a statement Wednesday.

The contrast between the White House's measured, vendor-neutral language and the Pentagon's harder line on Anthropic reflects the complexity of AI governance within the federal government, where civilian and defense priorities do not always align.

Anthropic, which develops the Claude family of AI models, has positioned itself as a safety-focused lab and has sought federal contracts. The specific events surrounding the Mythos situation — which appears to have precipitated the Pentagon's decision to distance itself from the company — were not elaborated on by Michael in his public remarks.

The Pentagon's stated shift toward a multi-model posture carries practical implications for how defense agencies will structure AI contracts going forward, potentially opening competition to a wider field of model providers rather than consolidating around a single vendor relationship.

Wiles's comments, meanwhile, suggest the administration is walking a careful line — emphasizing speed and safety in AI deployment while avoiding the appearance of directing government favor toward specific companies in an increasingly competitive industry.

For Anthropic, the Pentagon's public and categorical rejection of future collaboration represents a significant setback in its federal market ambitions, even as the broader White House stance leaves some room for the company's relationship with civilian agencies to continue on its current trajectory.