The US government has allowed Anthropic to restore access to its advanced Mythos 5 artificial intelligence model for a selected group of trusted organizations, partially reversing restrictions introduced earlier this month over national security concerns. Anthropic confirmed that more than 100 approved organizations, including many Fortune 500 companies and institutions, will regain access to the cybersecurity focused AI model following government approval. The decision comes after Anthropic was directed on June 12 to suspend access to its most advanced AI models, Mythos 5 and Fable 5, under export control measures intended to reduce potential national security risks. According to the company, Mythos 5 will now be made available to organizations responsible for operating and defending critical infrastructure, while discussions with government authorities continue regarding broader deployment of both models.
The latest decision reflects the US government’s continued oversight of advanced frontier AI models amid concerns that highly capable systems could be misused by foreign military or intelligence organizations. Anthropic stated that it had been informed that Mythos 5 could be redeployed to approved organizations, adding that it is working with authorities to expand access further and eventually restore public availability of Fable 5. Earlier on the same day, OpenAI also confirmed that it was limiting access to its GPT 5.6 models at the request of the US government, restricting deployment to a small group of vetted partners before a wider release. Both developments highlight the administration’s increased scrutiny of advanced AI technologies following executive actions introduced by U.S. President Donald Trump. Those measures established a voluntary framework allowing developers of covered frontier AI models to provide government agencies with early access for evaluation before broader distribution to trusted organizations.
The government’s role in determining which organizations receive access has generated criticism from legal experts and technology leaders. John Coleman, legislative counsel at the Foundation for Individual Rights and Expression, questioned the lack of transparency surrounding the approval process, arguing that limited public information about how organizations are selected raises concerns regarding oversight and fairness. OpenAI Chief Executive Officer Sam Altman also commented publicly that while extensive safety testing is appropriate, government involvement in selecting customers for advanced AI models is a matter of concern. Security specialists have noted that AI systems such as Mythos 5 possess advanced cybersecurity capabilities that could potentially accelerate sophisticated cyber operations if deployed irresponsibly, particularly against industries with complex and interconnected technology environments, including the banking sector. However, government officials have not publicly disclosed the specific safeguards introduced before approving the latest release.
According to a letter from US Commerce Secretary Howard Lutnick, Anthropic has made significant progress in addressing government concerns associated with covered AI models, allowing export licensing requirements to be removed for approved organizations and their non US citizen employees. Licensing restrictions will remain in place for organizations that are not included on the government’s approved list. Sources familiar with the matter indicated that many of the authorized organizations participate in Anthropic’s Project Glasswing, which includes approximately 100 prominent technology companies and institutions. Officials are also evaluating the future release of Fable 5, although no timeline has been announced. While both Mythos 5 and Fable 5 are built on the same underlying AI model, Mythos is intended for restricted cybersecurity applications whereas Fable is designed for broader public use. The latest policy changes follow ongoing discussions between AI developers and government agencies over balancing innovation, cybersecurity, export controls, and responsible deployment of increasingly capable artificial intelligence systems.
Follow the SPIN IDG WhatsApp Channel for updates across the Smart Pakistan Insights Network covering all of Pakistan’s technology ecosystem.
