U.S. Government Forces Anthropic to Pull Its Most Powerful AI Models in Unprecedented National Security Order

The U.S. government, citing national security authorities, issued an export control directive on June 12 ordering Anthropic to suspend all access to its Fable 5 and Mythos 5 AI models by any foreign national — whether inside or outside the United States, including foreign nationals employed by Anthropic itself. The result was immediate and sweeping: rather than attempt a selective enforcement the company had no technical means to implement, Anthropic disabled both models for every customer on earth.

A U.S. official confirmed that the Commerce Department sent the letter. Anthropic said it received the directive at 5:21 p.m. ET and that the letter did not provide specific details about the underlying national security concern.

This is not a routine compliance matter. The broad directive targeting Fable 5 and Mythos 5 represents one of the furthest-reaching actions the U.S. government has taken in response to the advanced capabilities of an AI model.

What Triggered the Order

Anthropic said it believed the government had become aware of a method of jailbreaking Fable 5 — a technique for bypassing its internal safety guardrails. The company reviewed a demonstration of the method and reached a very different conclusion than Washington did.

The vulnerabilities identified appeared relatively simple, and Anthropic found that other publicly available models were capable of discovering them as well — without any special technique. According to reporting by Bloomberg, the Commerce Department acted after another company claimed it had jailbroken Mythos. Anthropic said it had seen only verbal evidence of a narrow technique involving prompting the model to read a specific codebase and fix software flaws, and disputed that this justified pulling a commercial product used at scale.

In its public statement, Anthropic was pointed in its pushback. "The finding of a narrow potential jailbreak should not be cause for recalling a commercial model deployed to hundreds of millions of people," the company said, adding that any such action should run through "a statutory process that is transparent, fair, clear, and grounded in technical facts." It concluded bluntly: "This action does not adhere to those principles."

Despite the dispute, Anthropic said it would comply while working to restore access. The company described the suspension as likely a misunderstanding rather than a settled finding, and emphasized that all other Claude models remain available.

A Strained Relationship Gets Worse

The June 12 directive didn't emerge from nowhere. This is not Anthropic's first conflict with the Trump administration. In an earlier dispute with the Pentagon over the use of its models, the Department of Defense labeled Anthropic a supply chain risk — a designation that made it impossible for the company to obtain government contracts, though it did not stop Anthropic from selling to corporations.

The current order comes amid an already strained relationship. In early March 2026, the U.S. Department of Defense classified Anthropic as a "supply chain risk." The company has navigated that friction while continuing to grow its commercial base — but this latest action is categorically different. Pulling flagship products offline is a different order of magnitude from losing government contract eligibility.

Anthropic described the suspension as a misunderstanding and said it is working on restoring access. The company acknowledged that perfect jailbreak resistance is not achievable for any model, while simultaneously contradicting the assertion that a narrow potential jailbreak should justify recalling a commercial model deployed to hundreds of millions of people. If this standard were applied across the industry, it would amount to a halt on all new frontier model releases.

The Crypto Market Angle

For crypto investors, this story has an unusually direct line into their portfolios. On-chain traders have spent 2026 racing to bet on the year's biggest private companies before they go public, and Anthropic is among the most heavily wagered.

The reaction was swift. Following the announcement, the perpetual contract tracking Hyperliquid's pre-IPO pricing expectations for Anthropic dropped approximately 3.7%, as the market began reassessing the AI firm's listing prospects. That's a meaningful single-session move for a pre-IPO instrument, reflecting genuine uncertainty about how a government-imposed model suspension affects an eventual valuation.

Meanwhile, decentralized AI tokens moved sharply in the other direction. Bittensor's TAO rose 13.4%, Venice Token climbed 18%, and Internet Computer rose 9.8%. The logic is straightforward: Washington's ability to switch off one of the world's most-used AI systems overnight makes the case for censorship-resistant, decentralized AI infrastructure more concrete than any whitepaper could.

The first recall of a widely deployed commercial frontier model intensified crypto's existing narrative around decentralized AI — the argument that on-chain AI infrastructure, beyond any single government's jurisdiction, offers something centralized providers structurally cannot.

What Happens Next

What happens next will largely depend on whether the government formalizes its concern through the kind of transparent process Anthropic is demanding, or lifts the order as quickly as it imposed it. Until then, two of the most capable AI systems on the market are dark worldwide.

Both Anthropic and U.S. government representatives have said they want the issue resolved quickly and the models released back to users. That shared interest in a fast resolution is perhaps the most reassuring signal for markets, though it offers no specific timeline.

For infrastructure teams and developers, the more lasting takeaway may be that frontier model availability can become a compliance event, a vendor-risk event, and a production-continuity event simultaneously — with essentially no warning. Anthropic received the directive mid-afternoon on a Friday and had the models offline by evening.

The incident has prompted markets to focus on two key questions: whether the government's order will be rescinded or narrowed in scope, and whether similar restrictions could be extended to other model categories. For Anthropic, both factors could materially affect its future valuation and IPO timeline.

For now, Opus 4.8 remains available, and Anthropic has given no indication that its other products are at risk. But the episode has made clear that even the most widely deployed AI systems operate within a regulatory environment that can move faster than any terms of service or product roadmap.

Crypto traders watching the decentralized AI sector, pre-IPO contract markets, and any forthcoming Commerce Department guidance will have plenty to track in the days ahead.

Reporting contributed from Anthropic's official statement, Bloomberg, CNN Business, and Heise Online.