The chaos of human mythology is getting worse

It’s been two weeks since Anthropic retired its Mythos-class models following a Friday evening ultimatum from the Trump administration. The company immediately sprang into action, sending a barrage of executives to Washington, D.C. But updates have been suspiciously missing, with no solution in sight.

Anthropic declined to comment several times this week on the status of the talks, saying there was no news to share. But there is little news He is The story is here. After 14 days of intense negotiations, no one knows when or if Anthropic’s most powerful AI models will return, let alone whether President Trump can expand his request to include more companies with similar technology. The more days pass without any solution, the more dire things become — not just for humanity, but for the entire American AI industry.

The Trump administration’s June 12 export control order required Anthropic to suspend “any foreign national’s” access to Mythos 5 and Fable 5 due to security concerns. This prohibition includes any non-U.S. citizen inside or outside the United States, including people who work for Anthropic. So far, Anthropic has figured its only option is to keep these models offline.

It is not clear exactly why anthropology and management remain at an impasse. There may be one problem There is no clear framework To implement export controls on artificial intelligence systems. Most companies that make dual-use products—civilian systems with potential defense or military uses—can evaluate them using what is essentially a checklist during the manufacturing and production process. However, anthropology faces a complex bureaucracy trying to figure out how to apply its rules from first principles.

This export control process can typically last over months, if not years, and ends before the product reaches the market. But as Edge I mentioned previouslyThe US Department of Commerce appears to have tested Fable 5 before its release and has not raised any complaints. Anthropic has concluded that its models are safe to deploy, a source familiar with the negotiations said. It appears that the agency did not act until someone (reportedly) came forward Amazon CEO Andy Jassy) A method to apparently break Fable 5’s guardrails was flagged – at which point the entire process was squashed in a few days.

Katie Moussouris, founder and CEO of Luta Security, reviewed a report on the Fable 5 vulnerability at Anthropic’s request. She thinks it’s greatly exaggerated. In a Blog postMoussouris detailed how researchers broke down the guardrails that prevent Fable 5 from finding exploitable vulnerabilities, one of the most terrifying abilities of the untethered Mythos 5. The form will reject code review requests for “security reasons,” but will accept “fix this code” requests followed by manual prompts, which could theoretically lead to vulnerabilities being flagged that were not supposed to be disclosed.

However, in Moussouris’s eyes, this was not supposed to lead to such drastic government action, but is in fact an essential tool for AI encryption. “Defenders should be able to ask the AI ​​to fix errors in the file, explain why the fix is ​​important, and write tests that confirm the patch worked,” she wrote. “This is not guardrail crossing. It is the most valuable thing an AI model can do for defense security: perform the search, repair, and test operations of defenders that are run every day.”

Last week, Anthropic co-founder Tom Brown replaced CEO Dario Amodei in negotiations with the Trump administration, along with Sarah Heck, the company’s head of public policy. Wired I mentioned. But negotiations still appear slow, if any progress at all.

Whatever the reasons for the delay, it was a serious blow to Anthropic. Before the lengthy negotiations, Anthropic was seen as the rare AI company with a path to profitability. Its Mythos-class models, whose input tokens sell for twice the cost of the lower-powered Opus 4.8, were supposed to boost its revenues ahead of its next IPO. Mythos’ cybersecurity prowess appeared to lead to a thaw in relations with the Trump administration months later Legal and rhetorical combat.

Anthropic needs Mythos’ revenue to pay for all the computing it recently acquired, including a deal to pay SpaceX $15 billion annually for access to its data centers, as well as for its public image ahead of its IPO. Two of Anthropic’s largest current shareholders — Google and Amazon — have carefully tried to stay on Trump’s good side, so they probably won’t be happy either.

At the same time, slow negotiations have also created a power vacuum in the global AI market, not just because of the shutdown of Mythos, but because the US government has signaled a willingness to shut down American AI systems that it deems risky — and several American companies, including OpenAI, Google, and Microsoft, have models that may pose similar risks to Mythos. Countries started calling Non-American Amnesty International. said Alex Stamos, cybersecurity expert and chief product officer at Corridor Edge Last week“One of America’s heroes is being kneed down by the US government while we’re in a race with the Chinese. It’s just incredibly stupid.”

As days go by, the situation gets worse for these companies. Their models approach Mythos-level capabilities that could trigger an export control order — in fact, OpenAI’s GPT-5.5 Cyber Just won 5 Myths About Certain Standards and the Trump Administration It is said It has just asked OpenAI to delay the release of GPT-5.6 due to security concerns, with plans for the government to approve each client one by one. Anthropic and OpenAI’s IPO is approaching. Every day, China is moving forward in the artificial intelligence race.

Ironically, the administration’s order comes after months of pushing to dismantle AI Guarantees and Systems – It is one of the first comprehensive regulatory decisions made by President Trump. But a whole host of cybersecurity leaders did We gather to tell If you have to organize, this is not the way to do it. For all the Trump administration’s pledges to roll back AI regulation under Biden, it appears to have regained that ground in many ways, and more.