"Passing the Test" Successfully: Mythos 5 is granted approval by the US government, $90 billion AI unicorn faces a turning point before going public.

Anthropic PBC has obtained approval from the U.S. government to conditionally restore partial access to its high-performance AI model Mythos 5. Previously, the Trump administration had imposed strict restrictions on it due to concerns that the technology could pose a threat to national security. In a letter obtained by foreign media, U.S. Secretary of Commerce Howard Lutnick wrote to Anthropic's Chief Computing Officer, stating, "Anthropic has been working with the U.S. government to address the risks associated with regulated models." He wrote that the efforts have made "significant progress," and the model could be made available to "specific trusted partners." This clearance comes after a sudden escalation two weeks ago when the U.S. government abruptly banned Anthropic from providing access to Mythos 5 and its related model Fable 5 to foreign nationals, citing concerns that the security barriers could be circumvented. This directive forced the company to shut down global access to both advanced systems and triggered intensive negotiations to find a solution. Commerce Department spokesperson Beno Cas stated, "In just two weeks, we have made relentless efforts to ensure that the U.S. maintains its global AI leadership while effectively preserving national security." However, the letter from the Commerce Department last Friday did not mention any adjustments to the restrictions on the use of the Fable 5 model. In a statement, Anthropic said, "We have received notice from the U.S. government that Mythos 5 - our most powerful cybersecurity model - can be redeployed to a small group of network defenders and infrastructure providers. We are working to quickly configure access for approved suppliers and restore their Mythos 5 access. We are pleased with this progress and continue to work with the government to expand access to Mythos 5 and push for the re-release of Fable 5 to the public." It is not yet clear what specific measures Anthropic has taken to address government concerns about advanced models being "jailbroken" or bypassing security barriers. According to an informed source, the company is expected to continue discussions with the government this weekend to restore access to Fable 5 as soon as possible. As the discussions are not public, the source requested anonymity. The source also said that Anthropic is working with the government to develop a policy framework to address such situations in the future. Earlier, in a blog post announcing the disabling of the two systems, Anthropic publicly opposed the government's decision to implement export controls, stating that "if this standard is applied across the industry, we believe it will effectively halt the deployment of new models by all leading model providers." The blog post also warned that the company believes "no model provider can currently achieve perfect jailbreak defense." The U.S. government's decision to restrict the availability of Anthropic models has also triggered a chain reaction in the AI industry, particularly affecting competitor OpenAI. Under pressure from the Trump administration, the latter restricted the release of its new powerful model GPT-5.6. Last Friday, OpenAI stated that, at the government's request, it would first release a preview version to specific approved partners and gradually expand availability in the coming weeks. Wider Friction The confrontation over Mythos 5 and Fable 5 is the latest clash between Anthropic and the Trump administration. For months, Anthropic has been in constant dispute with the Pentagon over additional safeguards for military AI tools. After contract negotiations broke down, the U.S. Department of Defense listed the company as a supply chain risk source in March and is now seeking other AI suppliers for the military. Privately held Anthropic has long touted itself as a more responsible AI developer than its competitors. In April of this year, the company shared the Mythos model with a limited group of businesses and institutions for the first time and warned that due to its ability to discover network vulnerabilities, widespread distribution posed a high risk. Earlier this month, Anthropic released Fable 5 as the first public version of a Mythos-level model, with safeguards in place to restrict its full network capabilities. Fable 5 is prohibited from responding to certain types of queries, including those involving network security and biology. In these cases, Anthropic's Claude chat Siasun Robot & Automation routes responses through another model named Opus 4.8. The continued restriction on the Fable 5 model by the federal government indicates caution towards the widespread availability of these models despite the added restrictions implemented by Anthropic before their public release. Mythos 5 is a version of Fable 5 with some security protections removed, originally intended for a more limited group of institutions for cybersecurity purposes under the invitation-only "Project Glasswing." As of earlier this month, around 200 companies had been granted access to Mythos through this program, including Apple, Google (a subsidiary of Alphabet), Cisco Systems, Nvidia, and Microsoft. Industry Impact The directive banning Anthropic's latest models is the most significant intervention by the U.S. government in AI company operations to date, highlighting emerging concerns about the security of cutting-edge models. This new challenge comes just weeks after Anthropic secretly submitted its initial public offering, and the company's latest valuation has exceeded $900 billion. According to sources, recent negotiations involved several high-level executives at Anthropic, including co-founder Tom Brown, who recently met with Lutnick and other senior government officials. U.S. officials informed Anthropic that once security concerns were addressed, the restrictions would be lifted. As the discussions are not public, these sources requested anonymity. During negotiations with the Commerce Department regarding Fable 5 and Mythos 5, Anthropic CEO Dario Amodei took a more detached role, helping to reduce friction between the two parties, according to the sources.