The US government has allowed Anthropic to restore limited access to its powerful Claude Mythos 5 artificial intelligence model, easing restrictions that were imposed earlier this month over national security concerns. Even so, access remains tightly controlled and is available only to a select group of trusted American organisations, not the general public.
The move came after talks between Anthropic and US officials, following a temporary suspension of access for foreign nationals and a broader pause on rollout. Authorities had worried that such an advanced AI system could be misused for cyberattacks or accessed by hostile foreign actors if it was released without strong safeguards.
Under the revised arrangement, more than 100 approved organisations, including cybersecurity firms, critical infrastructure operators and selected Fortune 500 companies, will be able to use Mythos 5. Many of these groups are already part of Anthropic’s Project Glasswing, an initiative that uses advanced AI to find and fix software weaknesses before attackers can exploit them.
Anthropic said the model has shown strong results in spotting complex software flaws and improving cyber defences. The company believes that giving trusted security experts access to the technology can help strengthen digital safety while lowering the risk of misuse. It also said it will continue working closely with US authorities before expanding access further.
The controlled rollout comes as governments around the world step up scrutiny of powerful AI systems. The Trump administration has introduced new rules requiring frontier AI developers to coordinate with federal authorities before releasing highly capable models. Similar restrictions have also affected other leading AI companies building next-generation systems.
Also Read: YouTube Shorts adds cleaner viewing tools