Anthropic Meets With Commerce Department as Feud Deepens. Business Spending Data Shows Sales Rising Despite It.

Since the Trump administration issued its export control directive on June 13, forcing Anthropic to disable Fable 5 and Mythos 5 for all users worldwide, the company's executives have held continuous meetings with administration officials, according to Reuters.

The most recent session was a Commerce Department meeting on June 16, reported by Forbes. The proceedings are confidential, and neither Anthropic nor the Commerce Department confirmed which executives attended or what was discussed.

What the Government Says It Saw

The directive invoked national security concerns and export controls to bar any foreign national, including Anthropic's own non-U.S. employees, from accessing the two models. According to Anthropic's own June 13 blog post, the company believes the U.S. government was shown a working "jailbreak" of Fable 5, a method to bypass its safety guardrails and access the capabilities of the underlying Mythos model.

Anthropic admitted in that post that "perfect jailbreak resistance is not currently possible for any model provider." The company said its strategy was "defense in depth": detecting jailbreak attempts and making universal bypasses too costly to execute reliably.

Fable 5 had been publicly available for exactly three days — released June 10 — before it was pulled. Mythos 5 had been in limited release before that.

The Backstory the White House Isn't Explaining Publicly

The Washington Post, citing two White House officials who spoke anonymously, reported that the administration had been weighing sanctions on Anthropic for weeks before the June 13 directive. The dispute, according to those officials, shattered what was already fragile trust between the two parties.

That fragility has roots in March, when the Pentagon designated Anthropic a supply-chain risk, an unusual step that effectively barred the U.S. military from using its models for Defense Department work. Anthropic has challenged that designation in court. The Pentagon's designation followed Anthropic's refusal to allow the government to use its models for mass surveillance of Americans and fully autonomous weapons systems, according to TechCrunch.

The administration has not publicly specified which clause of export control law it invoked, and the exact legal authority behind the order remains unconfirmed in any of these sources. No formal sanctions have been announced as of June 16.

The Strongest Argument for the Government's Position

The national security concern here is not manufactured. A model Anthropic itself described as dangerous enough to restrict, one good enough at finding software vulnerabilities that the company marketed it with explicit warnings, was publicly accessible for three days before being yanked. If the government was genuinely shown a functional jailbreak that unlocked Mythos-level capabilities through the public Fable interface, restricting foreign access while that vulnerability is assessed is a defensible call. No one has publicly disputed that the jailbreak demonstration happened. The open question is whether a blanket foreign-national ban, which also locked out Anthropic's own international staff, was proportionate to the specific risk.

The Data That Complicates the Narrative

The government's pressure campaign appears to be making Anthropic more attractive to business customers, not less.

Ramp, the financial platform used by more than 70,000 businesses, published data showing Anthropic's share of AI subscription spending by businesses hit 41% in May, up 2.5 percentage points from April. That put it ahead of OpenAI for the first time, which held 39.5% — essentially flat month-over-month, according to TechCrunch.

Ramp's lead economist, Ara Kharazian, told TechCrunch directly: "If anything, it'll probably boost them. Anthropic's best month on record, as far as business adoption, was the month that the Department of Defense labeled them a supply-chain risk. There's a lot of aura that comes with your model specifically being named too dangerous to use."

Ramp's data is not granular enough to show how much revenue Anthropic loses while Fable 5 and Mythos 5 remain offline. That's a real financial hit, particularly given the timing.

The IPO Wrinkle

Anthropic filed confidential IPO paperwork, and separately raised $65 billion at a $965 billion valuation in a private funding round at the end of May, reportedly on the strength of its first profitable quarter, according to TechCrunch. Pulling your two flagship models weeks before a public offering is not ideal. Whether investors price the government conflict as a liability or, as Kharazian suggests, as a perverse marketing asset, is a question the IPO process will eventually force into the open.

The more pressing unresolved question, as of June 16, is whether Anthropic and the Commerce Department can agree on a technical fix. Some form of verified access restriction short of a total market withdrawal might satisfy the government's jailbreak concern without keeping the models dark indefinitely. No timeline for restoration has been announced by either party.