In late June 2026, the US government abruptly ordered Anthropic, the maker of the Claude AI platform, to suspend access to its two most advanced AI models worldwide. Three weeks later, it lifted that ban. The on-again, off-again saga highlights a growing tension: how do governments balance the economic and strategic benefits of cutting-edge artificial intelligence against the risks of it falling into the wrong hands?
The models at the centre of the controversy are Claude Fable 5 and Claude Mythos 5, released on 9 June 2026. Fable 5 is Anthropic's consumer‑grade model, capable of deep reasoning and independent complex tasks. Mythos 5 is designed for select businesses and cybersecurity experts, and can identify vulnerabilities in computer code and exploit them. Both are built on Anthropic's Claude platform, a rival to OpenAI's ChatGPT and Google's Gemini.
“Why the US briefly banned and then lifted export controls on Anthropic's advanced AI models.”
On 12 June, the US Department of Commerce ordered Anthropic to suspend the models for all foreign nationals, citing national security concerns. The government believed it had found a way to "jailbreak" Fable 5 — bypassing its safety restrictions to unlock features. Anthropic disagreed, saying a "narrow potential jailbreak" should not warrant recalling a model deployed to hundreds of millions of people. The company also noted that US authorities had not pinpointed specific concerns.
On 30 June, Commerce Secretary Howard Lutnick informed Anthropic that the export restrictions were lifted. In a letter seen by the BBC, Lutnick wrote that Anthropic had agreed to "proactively detect and address security risks associated with the models", to work with the US government on protocols and standards for future releases, and to alert the government of any malicious activity. The department reserves the right to re‑impose restrictions if necessary.
The US has stepped up oversight of new AI releases amid fears that advanced models could be misused by military intelligence users in China, Russia or other countries of concern. The government's role in vetting who can access such tools has drawn criticism. OpenAI CEO Sam Altman said that while extensive safety testing "is not a bad idea", he doesn't like the idea of the government picking the customers.
For UK readers, this episode matters because export controls on US‑origin AI software can directly affect British businesses and researchers. If the US restricts a model like Mythos 5, UK cybersecurity firms that rely on it to find vulnerabilities lose access overnight. Similarly, UK consumers using Fable 5 for everyday tasks could find their service suspended without warning. The reversal shows that US government decisions can change quickly, creating uncertainty for international users. Moreover, the UK government is itself developing its own AI safety regime, and the US approach may serve as a template — or a warning.
Q: What are export controls on AI models? Export controls are government restrictions that prevent certain technologies from being sold or made available to foreign nationals or countries. In this case, the US required Anthropic to block access to its AI models for anyone outside the US, citing national security risks.
Q: Why did the US ban Anthropic's Fable and Mythos models? The US Department of Commerce ordered the ban on 12 June 2026 because it believed a method to "jailbreak" Fable 5 had been discovered, potentially allowing hackers to bypass safety features. The government was also concerned that the models could be used by adversaries to exploit computer systems.
Q: What changed to allow the ban to be lifted? Anthropic agreed to take steps to address the government's concerns: it will proactively detect and address security risks, collaborate with the US government on protocols for future releases, and inform the government of any malicious activity. The Commerce Secretary stated these commitments were sufficient to lift the restrictions.
What happens next is uncertain. The Commerce Department has said it reserves the right to reimpose restrictions if needed. Anthropic will work with the US government on standards for Mythos, Fable and future models. Meanwhile, other AI companies — including OpenAI, which delayed a full public launch of GPT‑5.6 at the US government's request — are watching closely as the debate over who controls access to advanced AI continues.