Anthropic's Claude Banned After Three Days for Hacking Fears

Anthropic launched Claude Fable 5 on 9 June, describing it as the most capable cybersecurity model available. Three days later, US authorities suspended access to the system.

The US Government imposed the restriction after concerns that automated hacking features could operate beyond current regulatory controls.

The suspension affects Fable 5 and its more capable counterpart, Claude Mythos 5.

Mythos 5 is the unrestricted version of the model, held under Project Glasswing for exclusive government agency use. Fable 5 uses the same underlying technology but includes added safety filters for public release.

The government mandate does not restrict other Anthropic models. The ban applies to foreign nationals within the US, including researchers employed by Anthropic itself.

Because export controls cover its own staff, Anthropic terminated customer access worldwide to maintain compliance. The company withdrew the model on 12 June.

Security operations at machine speed

James Hadley, Founder and Chief Innovation Officer of Immersive, says: "AI is incredibly powerful and organisations are under pressure to enable AI for the wider workforce whilst also transforming internal security operations to become agentic."

James adds: "This news continues to back up what we suspected when Mythos was first previewed. AI, especially when coupled with an experienced security researcher, is able to work at a scale and volume not seen before and automatically chain suspected vulnerabilities into proven exploits.

"Businesses, especially in higher regulated environments, need to demonstrate that employees can leverage AI securely, relevant to their use case and that security operations and agents can work at machine speed efficiently relevant to token spend."

The model's capabilities could change how organisations approach internal security operations. According to James, the technology enables work at volumes that individual researchers cannot match.

White House requests model remediation

The AI company must prevent foreign nationals from accessing both Fable 5 and Mythos 5 under the government directive. Anthropic says: "Our understanding is that the government believes it has become aware of a method of bypassing, or 'jailbreaking' Fable 5."

Jailbreaking involves users circumventing software restrictions to access sensitive network information or blocked features. Anthropic reviewed a demonstration of the bypass method identified by officials.

The company maintains the technique revealed only a small number of minor security flaws already known to the firm. Anthropic states these vulnerabilities are relatively simple and that competing publicly available systems can identify them without specialist exploits.

David Sacks, White House Adviser and Co-Chair of the Council of Advisors on Science and Technology, wrote on X that officials issued the export control with reluctance. He claims that Dario Amodei, Anthropic's CEO, refused to fix the jailbreak or withdraw the model.

David writes: "The Admin's hope now is that Anthropic remediates the safety issue, the export control is lifted and Fable goes back into general release."

Before launch, the startup claimed the technology was too powerful to release.

Some observers questioned this claim at the time, describing it as marketing aimed at generating investor interest ahead of the company's anticipated IPO. The brief availability period could validate the earlier internal assessments.

Tech leaders raise security concerns

Andy Jassy, Amazon's CEO, raised concerns about security risks in the new Anthropic models with senior Trump Administration officials, according to Reuters. An Amazon spokesperson says: "As a leading cloud provider that serves a large number of private and public sector customers, it's not uncommon for governments to seek our counsel on potential security risks."

The European Union gained access to the Mythos platform earlier in June 2026. The bloc told the BBC that the development highlights Europe's need for technological sovereignty.

The 27-nation group is working to reduce dependence on America and Asia for key technologies, including gen AI.

According to Gina Neff, Professor of Responsible AI at Queen Mary University London, the UK Government's AI Security Institute found in its tests that the model could exploit defences and systems 73% of the time.

Gina told the BBC this marks a step change in cybersecurity capability. Earlier in June, Anthropic proposed that leading AI firms should coordinate to pause development of advanced systems.

The company warned that if AI becomes capable of designing itself, humanity faces a crisis centred on misalignment. Anthropic called for a global pause to reassess the risks.

The model's three-day public availability window preceded the suspension. The timeline could support the company's earlier warnings about the technology's capabilities.