US lifts ban on Anthropic AI models after security fears

By the Editorial Team · Thursday, July 2, 2026

Anthropic has reopened access to its AI model Fable after a US government suspension over security concerns related to potential cyber-attack misuse. This decision follows enhanced safeguards and government collaboration, marking a shift toward tighter regulatory oversight of advanced AI technologies to mitigate national security risks.

Anthropic has resumed customer access to its advanced artificial intelligence model, Fable, following a suspension imposed by the US government over national security concerns. The restriction, which lasted more than two weeks, stemmed from fears that the model could be exploited to facilitate serious cyber-attacks.

The San Francisco-based company confirmed that export controls were lifted after collaboration with US officials. Commerce Secretary Howard Lutnick noted in a social media statement that authorities had worked closely with Anthropic “to analyze and approve Fable 5 to ensure alignment across the US Government and strengthen America’s leadership in AI.”

The initial ban came shortly after the June 9 launch of Claude Fable 5. The US government ordered Anthropic to immediately limit access to Fable 5 and the related Claude Mythos 5 model for foreign nationals both domestically and abroad. Officials expressed concern that these AI tools, due to their capabilities, might be used by adversaries to conduct cyber operations against critical infrastructure.

Last Friday, the government allowed restricted release of Mythos 5 to a limited group of “trusted” US organizations, with its usage focused on enhancing defensive cybersecurity measures. Lutnick, in a letter to Anthropic seen by Reuters, said the company had agreed to proactively monitor and mitigate security risks linked to its models, to cooperate extensively with the government on protocols governing current and future releases, and to report any malicious activity detected.

This move represents a marked shift in US AI policy, signaling a more cautious regulatory approach following earlier lighter oversight intended to maintain a competitive edge over China in AI development.

The US action was prompted by findings from Amazon researchers who demonstrated a method to circumvent Fable 5’s built-in safeguards, revealing how the model could be manipulated to identify software vulnerabilities — vulnerabilities that could potentially be exploited by hostile actors.

In response to the imposed export controls, Anthropic temporarily halted access to its models but has since enhanced security measures. The company stated in a blog post that researchers from the Commerce Department’s Center for AI Standards and Innovation (CAISI) tested both previous and updated safeguards and found them to be “extraordinarily strong.”

The development comes amid heightened warnings from intelligence agencies across Australia, the US, the UK, New Zealand, and Canada. These agencies cautioned that while AI holds promise for improving cyber defense capabilities, it also accelerates the pace, scale, and complexity of cyber threats.

The US government’s selective vetting of companies authorized to access advanced AI has drawn criticism within the industry. Sam Altman, CEO of OpenAI, urged caution regarding government involvement, stating that extensive safety testing is prudent but expressing concern over the government’s role in choosing which customers receive access. OpenAI, the creator of ChatGPT, has also delayed the broader release of its GPT-5.6 model at the government’s request, restricting it to a narrow circle of vetted partners.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.