Anthropic Forced to Pull Fable 5 AI Model After White House Export Ban

On Friday, June 13, Anthropic abruptly disabled its two newest and most powerful AI models — Fable 5 and Mythos 5 — for every single customer worldwide. The reason: an export control directive from the Trump administration, issued with just 90 minutes to comply, that barred foreign nationals from accessing the models entirely.

According to The Hill, the order came after Amazon CEO Andy Jassy personally called the White House to flag a jailbreak vulnerability his researchers had discovered in Fable 5's cybersecurity guardrails. The Commerce Department, led by Secretary Howard Lutnick, then issued the export control order citing "national security authorities."

"The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance," Anthropic,² according to CIO. The company couldn't distinguish between US citizens and foreign nationals in its user base, so it had to pull the plug for everyone.

The dispute centers on whether the jailbreak Amazon discovered actually represents a genuine national security threat — or whether it's a narrow vulnerability that exists in every frontier model.

³ tested Fable 5 before it was pulled and reported it was "an impressively intelligent AI model, even compared with GPT‑5.5 and Opus 4.8." The jailbreak in question was a "non‑universal" bypass — meaning it allowed users to ask Fable 5 to read a specific codebase and find software flaws, but didn't broadly defeat all of the model's safeguards.

Anthropic pushed back hard. In a blog post, the company argued that PCMag reported "the outlined vulnerabilities are neither new nor significant" — and noted that other publicly available models, including OpenAI's GPT‑5.5, could discover the same vulnerabilities without requiring a jailbreak at all.

The Hill reported that Anthropic's blog post went further: "If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers."

More than 100 cybersecurity and tech policy experts signed an open letter arguing that Fable and Mythos were essential tools for cyber defenders to find and patch vulnerabilities in their own systems, and that these benefits outweighed the risk of attackers jailbreaking Fable, Fortune reported.

The ban triggered a scramble behind the scenes. Anthropic raced to send senior leaders with research and safeguard expertise to Washington D.C. for the first in‑person meetings since the restrictions took effect.

According to Business Insider, the company's delegation included Logan Graham from the Frontier Red Team, Dave Orr (head of safeguards), and Nicholas Carlini (lead security researcher). They gave a presentation to administration officials explaining Anthropic's cybersecurity safeguards in hopes of resolving the restrictions.

The weekend before involved multiple hours‑long calls between Anthropic co‑founder Tom Brown, Commerce Secretary Howard Lutnick, and National Cyber Director Sean Cairncross, Business Insider reported. Treasury Secretary Scott Bessent, who had been involved in earlier discussions with Anthropic CEO Dario Amodei, was less involved over the weekend as Lutnick took the lead.

A senior White House official told ⁵ that resolving the restrictions would likely take longer than a few days, but left the door open to a faster resolution: "That's up to Anthropic."

The most immediate market reaction to the Anthropic ban came from an unexpected direction: Chinese open‑source AI companies surged as investors bet the export controls would accelerate demand for models companies can run on their own infrastructure.

CNBC reported that MiniMax and Zhipu, the Chinese open‑source AI lab, both surged on Monday. The calculus is straightforward: an open‑source model can be downloaded, run on a company's own servers, and customized for its data — and no government order can switch it off.

Microsoft CEO Satya Nadella, whose company is the principal backer of OpenAI and also invested billions in Anthropic, weighed in on X. "The last thing any of us want is a world where every company across every sector is ceding value to a few models that eat everything they see," Nadella wrote, a message ⁶ characterized as a warning about the risks of closed‑model dependency.

The shift toward self‑hosted models is accelerating. Yash Patel told ⁶ that the Anthropic fight "highlighted the significance of owning your own model."

For the thousands of developers who adopted Fable 5 in its first week, the shutdown is a hard lesson in platform risk. Your entire product can be yanked offline overnight by a government order you never saw coming — and there's nothing you can do about it.

The export ban exposes a fundamental vulnerability in the closed‑model ecosystem. When you build on someone else's API, you're not just betting on their uptime and pricing — you're betting on the geopolitical landscape. CNBC noted that this "drove home a hard truth for the companies that were counting on them: access can be cut off at any time."

There's a practical near‑term fix for US‑based consumers. CIO reported that Anthropic's updated privacy policy, set to take effect July 8, introduces a new identity verification provision. Users could submit a government ID scan to prove US citizenship and regain access to Fable 5 — but enterprise customers with international teams don't have that option.

The broader significance of the Anthropic ban extends far beyond one company. As ⁴ put it, the U.S. government has effectively created "a licensing regime for frontier AI — it just doesn't want to admit it." The process was ad hoc and opaque: a single CEO's phone call triggered a chain of events that took a model offline within days, with no public hearing, no published standard, and no clear appeals process.

The fallout is already global. European governments are now scrambling to develop sovereign AI capabilities in the wake of the ban, Fortune reported, with policymakers asking whether Washington can pull the plug on AI sales abroad whenever it chooses. Politico reported that the White House's move has "jolted Congress back into the AI debate," with Democrats and Republicans remaining divided over whether advanced AI models should be vetted before public release.

As of Tuesday, June 16, Anthropic and the Trump administration were still in negotiations. The company's models remain offline, and no timeline for resolution has been announced.

Several outcomes are possible. The most optimistic: Anthropic convinces the administration that its safeguards are adequate, and Fable 5 returns with some form of geographic access controls. The middle path: a prolonged negotiation that keeps the models offline for weeks. The worst case: the ban becomes a template for future export controls on frontier AI, creating a permanent two‑tier system where the most capable models are unavailable to anyone outside the US.

For builders, the message is clear. The era of assuming your AI tools will always be available is over. Diversify your model providers, invest in open‑source alternatives, and build your stack so that no single government order can take your product offline. The Anthropic ban isn't just about one company — it's the new reality of building with AI.