The US Department of Commerce has lifted the export controls that forced Anthropic to freeze foreign access to its two most advanced AI models, Claude Fable 5 and Mythos 5. Commerce Secretary Howard Lutnick announced the reversal on June 30, 2026, ending a roughly two-week standoff between the company and the Trump administration, with global access beginning to be restored from July 1.
Key Highlights
- Export controls on Claude Fable 5 and Mythos 5 were lifted on June 30, 2026, less than a month after they were imposed.
- The original restrictions had blocked all foreign nationals — including non-citizen Anthropic employees, whether inside or outside the US — from using the models.
- Anthropic agreed to proactively detect security risks, collaborate on standards for future models, and report malicious activity to the government.
- Fable 5 returned to global users across the Claude platform, Claude.ai, and Claude Code, with full restoration rolling out from July 2.
Details
The controls were imposed shortly after Anthropic publicly released Fable 5 on June 9, 2026, a safeguarded version of its flagship Mythos model. The White House abruptly restricted access after the discovery of what Anthropic described as a "jailbreak" method affecting Fable 5. The company downplayed the finding, characterizing it as exposing "minor vulnerabilities" rather than enabling sophisticated cyberattacks.
Mythos 5 is Anthropic's most advanced model, initially offered only to select technology companies, governments, and financial institutions, and is capable of identifying cybersecurity vulnerabilities. Fable 5 is the publicly released version of Mythos, wrapped in additional safeguards that restrict cybersecurity, biology, and chemistry-related queries.
Secretary Lutnick said the administration had "worked closely" with Anthropic to "analyze and approve Fable 5" and to strengthen "America's leadership in AI." White House Chief of Staff Susie Wiles framed the shared objective as getting "the best tech deployed as quickly and safely as possible." Anthropic, for its part, said: "We're grateful to our users for their patience, and to everyone who worked with us on redeploying the models."
Impact
The reversal signals that the US government is willing to walk back sweeping restrictions once a frontier lab demonstrates security cooperation — but Lutnick's official letter warned that controls could be reinstated if "circumstances change" or if Anthropic fails to meet its commitments. That leaves frontier model access explicitly conditional on ongoing government approval rather than guaranteed.
For developers and enterprises across the MENA region — including Tunisia, Saudi Arabia, and the Gulf — the lift is immediate good news: foreign nationals can once again use Fable 5 on Claude.ai and Claude Code. But the reinstatement clause underscores that access to the most capable models now depends on geopolitics and governance, not just pricing and API keys. Teams building on frontier capability may still want a self-hostable, open-weight fallback as a second provider to hedge against sudden availability shocks.
Background
The episode is the latest chapter in a tense relationship between Anthropic and Washington. In March 2026, Anthropic sued the Department of Defense over a "supply chain risk" designation. Anthropic met with the Trump administration on June 15 as tensions escalated, and by June 27 had secured approval for critical infrastructure organizations before the broader lift days later.
The move mirrors a wider pattern of government gatekeeping over frontier AI. In the same window, OpenAI released GPT-5.6 only to a small set of government-vetted partners under a staggered rollout requested by the White House.
What's Next
Anthropic will continue restoring full global availability through early July. The company has committed to collaborating with the government on security standards for future models, meaning the approval process could become a recurring feature of each frontier release. Industry observers, including Sophont's Tanishq Abraham, called the reversal a "big deal" that raises precedent questions for how AI models are regulated going forward, while Francesco Bailo of the University of Sydney suggested the government "likely realised it had overreacted."
Source: CNBC