Damage control? Anthropic rushes to Washington amid White House ban on top AI models ahead billion-dollar IPO
Anthropic sent its top technical staff to Washington for face-to-face meetings with White House officials as the company scrambles to resolve a deepening dispute that has taken its most advanced AI models offline for users around the world, a person familiar with the development said. axios.
How did the export control crisis arise?
The crisis began on June 12, when the U.S. Department of Commerce issued an export control directive ordering Anthropic to deny foreign nationals access to its two newest models, Fable 5 and Mythos 5, both within the United States and abroad. Since it is not technically possible to reliably filter users by nationality, Anthropic has made the decision to completely disable both models for all users worldwide.
The closure comes just three days after the models were made available to the public on June 9. Luck The move, described as a “major step” for the lab, comes a little more than a week after Anthropic secretly filed for IPO paperwork.
What are Myths and Fairy Tales and why are they important?
Mythos is Anthropic’s most powerful frontier model. When the company first announced this in April, it acknowledged that the model was too proficient in cybersecurity to be released safely and touted its significantly improved ability to identify vulnerabilities in software.
Rather than a public distribution, Mythos was made available to a small number of trusted organizations, mostly US technology companies, to help identify and remediate vulnerabilities in critical digital infrastructure.
Fable 5 is built on the same basic model but includes additional security measures designed to prevent it from being used for cybersecurity purposes. It was this version that Anthropic made public on June 9 and almost immediately shut down under pressure from the US government.
The hidden constraint that sparked the backlash
Within hours of Fable 5’s public release, significant criticism began to circulate among AI researchers, developers, and policy experts. The backlash focused on a statement embedded in the model’s 319-page system board, a document that provides detailed security information, revealing that Fable 5 would silently drop its own responses when it detects queries related to advanced AI development work, such as building the infrastructure used to train large AI models.
In practice, this meant that the user could ask the model a question, receive a deliberately attenuated response, and have no indication that the model was hiding anything.
Unlike Fable’s other restrictions, which cover cybersecurity and biology and visibly redirect users to a less capable model with clear notification, the system card stated that this particular limitation “is not visible to the user.” The model would still respond, but would implement “interventions that would limit Claude’s effectiveness” without any explanation to the questioner.
Anthropic estimated that the restrictions would affect roughly 0.03 percent of traffic and defended the approach, stating that “enforcing this restriction through our security measures prevents the actors most willing to violate these terms from stepping up.”
Anthropic and the Trump administration: an escalating conflict
Export controls did not emerge from a vacuum. Since early 2025, relations between Anthropic and the Trump administration have been increasingly deteriorating. Administration officials accused the company of producing “woke artificial intelligence” and called CEO Dario Amodei an “ideological lunatic.”
Initial tensions centered on disagreements over AI regulation and semiconductor export policy. The conflict intensified further when Anthropic refused to allow the Pentagon to use its models for domestic surveillance operations and fully autonomous weapons systems. The Department of Defense responded by threatening to designate Anthropic as a “supply chain risk,” a classification that would force military contractors to cut ties with the company altogether.
Talks continue but trust remains fragile
Despite the seriousness of the situation, management officials claimed Anthropic failed to take serious action. But a source close to the company said Anthropic’s technical staff has been holding virtual meetings with White House officials since the administration made its initial approach on Friday, and that the personal visit to Washington represents a significant escalation in the company’s efforts to find a solution.
The risks are high. With its IPO filing underway and its flagship models on hold globally, Anthropic faces increasing pressure to demonstrate that it can simultaneously navigate the intersection of AI security, government regulations, and commercial viability.
