A two week standoff between a software company and the United States government ended on Friday with a deal that neither side would have chosen at the start.
Anthropic got some of what it wanted. The government kept control of most of it. And millions of paying customers around the world remain locked out with no clear date for when that will change.
Here is what actually happened and what it means. On the 13th of June, the United States government ordered Anthropic to shut down access to its two most advanced programmes. One was Mythos 5, a tool built specifically for identifying and defending against weaknesses in computer systems. The other was Fable 5, a general purpose tool that had been made available to the public just days before the ban was imposed. Both were pulled offline at short notice. Users worldwide lost access without warning.
The order was enforced through export control rules. These are laws that govern which technologies can be shared with foreign nationals and foreign countries. The government’s concern, as reported at the time, was twofold. First, officials had been told about a method for bypassing the protective filters built into Fable 5. Second, there were concerns inside the White House that foreign actors including those connected to China might use Mythos 5 to attack American systems.
For fourteen days, Anthropic and the government held intensive daily talks. The company wanted its programmes back online. The government wanted assurances that appropriate safeguards were in place. On Friday evening, those talks produced a written agreement.
Commerce Secretary Howard Lutnick wrote a letter to Anthropic co-founder and chief compute officer Tom Brown. The letter stated that appropriate safeguards were now in place. It granted permission for Mythos 5 to be made available again. But the permission came with strict conditions.
Only approved organisations can access the programme. The list covers more than 100 entities. These include major American companies and federal government agencies. Foreign nationals employed by those approved organisations are also covered. So are Anthropic’s own staff who are not American citizens.
Everyone else remains shut out. Companies and individuals not on the approved list cannot access Mythos 5. And Fable 5, the consumer-facing programme used by millions, was not mentioned in the letter at all. People close to the talks told reporters that discussions about Fable 5 were continuing and moving in a positive direction. No timeline was given.
For Anthropic, Friday’s outcome is a partial win. The company can now serve the institutions that arguably need Mythos 5 most. These include organisations that defend critical computer networks. But the commercial market remains frozen. Subscribers who were paying for access have been waiting two weeks and are no closer to a confirmed return date.
The background to this dispute runs deeper than the past fortnight. As reported, the tension between Anthropic and the current American administration began months earlier. In February this year, President Donald Trump signed an order telling all federal agencies to stop using Anthropic products. That order followed Anthropic’s refusal to allow its programmes to be used for certain military applications and for surveillance of the American public at scale. The Pentagon then placed Anthropic on a supply chain risk list. That designation effectively barred military contractors from working with the company. Anthropic challenged that decision in court and the case is still being heard.
Against that backdrop, the June shutdown landed at a particularly difficult moment for the company. It had just launched Fable 5 to widespread attention. It was weeks away from confidentially filing paperwork for a public stock market listing. And it found itself in the unusual position of being simultaneously engaged in legal action against the government and daily negotiations with it.
See Also: OpenAI limits GPT-5.6 model access to trusted partners after US government request
The Friday letter represents the government’s clearest signal yet that the relationship is thawing. Commerce Secretary Lutnick’s willingness to write directly to Anthropic and certify that safeguards are adequate is a meaningful step. It does not resolve the underlying disputes about military use and surveillance. Those remain contested in court. But it separates the commercial access question from the broader policy fight, at least for now.
OpenAI, Anthropic’s closest rival, chose the same Friday evening to announce three new programmes of its own. The company said it was releasing them to a small group of government-approved trusted partners first, with a broader rollout planned for the coming weeks. It described the approach as something it believed in wholeheartedly. That framing drew immediate contrast with Anthropic’s situation, where access to programmes was not staged by choice but removed by order.
Both companies are navigating the same new reality. The most capable software tools in existence are no longer treated as ordinary commercial products. They are treated as strategic assets. Who gets to use them, and on what terms, is now partly a government decision.
About the Author
Marcel Chidozie is a tech analyst and writer covering foreign news, fintech, and emerging technologies at TechRegard. Based in Nigeria, He's passionate about translating complex tech developments into compelling, accessible stories for diverse audiences. His work focuses on how technology shapes innovation across Africa and globally.
