MUMBAI: The AI gates have creaked open again but only for a select few. Anthropic has begun restoring access to its advanced artificial intelligence model, Claude Mythos 5, after the US government partially relaxed restrictions that had temporarily blocked its release over national security concerns.
The company said it has received approval to make Mythos 5 available to a select group of trusted US organisations responsible for operating and protecting critical infrastructure. According to reports, more than 100 organisations, including several Fortune 500 companies, will regain access under the revised directive.
The decision partially reverses a government order issued on 12 June, which required Anthropic to suspend access to its frontier AI models Mythos 5 and Fable 5 amid concerns that highly capable AI systems could be exploited by hostile foreign actors.
Announcing the development, Anthropic said the government had authorised the redeployment of Mythos 5, describing it as the company’s most advanced cybersecurity-focused model. The firm added that it is working to restore access quickly while continuing discussions with US authorities to broaden availability and eventually resume public access to Fable 5.
The latest move comes as the US government tightens oversight of frontier AI systems. Rival OpenAI also confirmed that it had delayed the wider rollout of GPT-5.6 at the government’s request, restricting early access to a limited group of vetted partners.
The government’s growing role in determining who can use advanced AI models has sparked debate across the technology industry. Critics argue that allowing authorities to decide which organisations receive access raises questions around transparency, competition and innovation.
OpenAI Chief Executive Sam Altman acknowledged the importance of safety testing but voiced concern over the government effectively deciding which customers can access cutting-edge AI technologies.
Anthropic said the revised directive also removes the requirement for export licences when trusted organisations or Anthropic employees who are not US citizens—access Mythos 5. However, licensing requirements will continue to apply to organisations outside the approved group.
Many of the organisations receiving access are understood to be participants in Project Glasswing, Anthropic’s initiative involving around 100 technology companies and institutions. The US government is also considering broader approval for Fable 5, although no timeline has been announced.
The restrictions stem from an executive order signed earlier this month by US President Donald Trump, creating a voluntary framework under which developers submit frontier AI models for government review before releasing them to trusted partners. The latest easing signals a calibrated approach, one that seeks to balance national security concerns with the growing demand for advanced AI capabilities.
