Deutsch
English
Español
Français
Nederlands
Português
Anthropic's Fable 5 Blocked by US Export Control Order
AI company Anthropic has been locked in days of unsuccessful negotiations with the US government to lift a global ban on its most advanced AI model, Claude Fable 5. The emergency export control directive, issued on June 12, 2026, by Commerce Secretary Howard Lutnick, ordered Anthropic to immediately suspend access for all foreign nationals—effectively forcing the company to disable the model worldwide. The unprecedented action marks the first time the US has used export control authority to recall a commercial AI deployment, and it has sent shockwaves through the tech industry and allied governments alike.
Background: From Pentagon Deal to Supply Chain Risk
The Fable 5 ban did not emerge in a vacuum. Tensions between Anthropic and the Trump administration escalated after the company refused Pentagon demands in February 2026 to remove contractual restrictions prohibiting the use of its AI for domestic surveillance and fully autonomous weapons. In response, the Department of Defense designated Anthropic a supply chain risk, barring military contractors from doing business with the firm—a designation Anthropic successfully challenged in court with a temporary injunction in March 2026.
The Anthropic US government dispute set the stage for the current crisis. When Anthropic launched Fable 5 on June 9, 2026, it was hailed as the company's most capable public model, built on the previously restricted Mythos architecture. The model excelled at identifying cybersecurity vulnerabilities—a feature that would prove to be its undoing.
The Jailbreak That Triggered the Ban
Amazon's Alleged Discovery
According to multiple sources, the trigger for the export control order was a demonstration by Amazon, a major Anthropic investor, of a potential jailbreak method. The technique allegedly involved asking Fable 5 to read a specific codebase and identify software flaws—capabilities that Anthropic says are widely available in competing models like OpenAI's GPT-5.5. The US government concluded there was an 'unacceptable risk' that adversaries such as Russia and China could exploit Fable 5 for cyberattacks.
Anthropic strongly disputes the severity of the vulnerability. In a public statement, the company described it as a 'narrow, non-universal jailbreak' that revealed only minor, previously known vulnerabilities. The company noted that it had implemented a defense-in-depth strategy, including routing dangerous requests in cybersecurity, biology, and chemistry to the less capable Claude Opus 4.8—occurring in fewer than 5% of sessions.
Global Shutdown
Because Anthropic cannot verify the nationality of every API caller in real time, compliance with the directive meant disabling Fable 5 and its sibling model Mythos 5 for all users worldwide—including US citizens and companies. The decision impacted hundreds of millions of users and countless businesses that had integrated Fable 5 into their workflows. Other Claude models—Opus 4.8, Sonnet 4.6, and Haiku 4.5—remain operational.
Anthropic CEO Dario Amodei, who ironically had published a policy essay just two days before the ban advocating for government authority to block unsafe AI deployments, found himself on the receiving end of that authority. The company's engineers flew to Washington for in-person negotiations over the weekend, but as of June 17, the ban remains in place.
G7 Summit: AI Governance Takes Center Stage
The controversy unfolded against the backdrop of the G7 summit in France, where AI governance emerged as a central theme. Amodei joined OpenAI's Sam Altman and Google DeepMind's Demis Hassabis as informal advisors to world leaders. French President Emmanuel Macron urged the US to share cutting-edge AI technology and called for democratic cooperation on AI regulation.
President Trump told reporters that negotiations with Anthropic were 'going fine,' but the UK's request for an exemption to access Fable 5 was formally denied, with a Trump administration official calling exemptions 'completely illogical.' The G7 AI regulation summit highlighted growing tensions between US export controls and allied nations' desire for access to frontier AI capabilities.
Europe's Wake-Up Call on AI Sovereignty
EU Technology Commissioner Henna Virkkunen used the Fable 5 ban to underscore Europe's vulnerability. Speaking at the G7, she warned that Europe must not remain technically dependent on the US or China for advanced AI. The best AI programs for detecting cybersecurity vulnerabilities come from either the US or China, with no European alternatives at the same capability level.
Just weeks before the G7, the European Commission unveiled a major tech sovereignty package, including the Cloud and AI Development Act (CADA) and Chips Act 2.0, aimed at tripling data center capacity within five to seven years and boosting domestic semiconductor manufacturing. The Fable 5 ban provides a concrete example of the 'kill switch' risk that European policymakers have warned about—where a foreign government can unilaterally cut off access to critical AI infrastructure.
Experts in European AI sovereignty suggest the ban could accelerate EU efforts to develop independent AI capabilities, though significant technical and financial hurdles remain.
Impact on the AI Industry
The Fable 5 ban sets a far-reaching precedent. For the first time, export controls have been applied not to hardware like semiconductors, but to a deployed commercial AI model based on user citizenship. Industry observers warn that if this standard is applied broadly, it could halt all new frontier model deployments across the industry.
Anthropic's statement captured the stakes: 'We strongly believe that applying this standard across the industry would essentially halt all new model deployments for all frontier model providers.' The company called for a transparent, fair, and technically-grounded statutory process for such decisions.
Businesses and developers that had built their workflows around Fable 5 lost access overnight. Stripe had reportedly used Fable 5 to complete a codebase-wide migration of a 50-million-line Ruby repository in a single day—a capability now unavailable. The broader economic impact remains to be seen, but the incident has already sparked debate about the balance between national security and technological progress.
Frequently Asked Questions
What is Claude Fable 5?
Claude Fable 5 is Anthropic's most advanced AI model, built on the Mythos architecture. It was designed to excel at software engineering, scientific research, and identifying cybersecurity vulnerabilities, with safeguards to prevent misuse in high-risk domains.
Why did the US government ban Fable 5?
The US Commerce Department issued an emergency export control directive on June 12, 2026, citing national security concerns over a potential jailbreak method that could allow adversaries to use Fable 5 for cyberattacks. The order required Anthropic to block access for all foreign nationals.
Is Fable 5 still available?
No. Anthropic disabled Fable 5 and Mythos 5 for all users worldwide because the company cannot verify user nationalities in real time. Other Claude models (Opus 4.8, Sonnet 4.6, Haiku 4.5) remain accessible.
How does this affect European AI development?
The ban has intensified European calls for AI sovereignty. EU Commissioner Virkkunen warned that Europe cannot depend on the US or China for critical AI capabilities, and the incident is likely to accelerate EU investments in domestic AI infrastructure under the new tech sovereignty package.
What happens next for Anthropic?
Anthropic continues negotiations with the US government to restore access. CEO Dario Amodei is at the G7 summit in France, where AI governance discussions are ongoing. The company has called for a transparent regulatory process and is working to propose technical solutions that satisfy national security concerns.
Follow Discussion