Negotiations between the White House and Anthropic concerning AI model restrictions appear to be improving, with leadership in the talks shifting away from CEO Dario Amodei. According to a report published on June 25, 2026, discussions are now directed by Anthropic co-founder Tom Brown and head of public policy Sarah Heck. Sources familiar with the process indicated that officials find the new team more approachable and effective at engagement compared to previous direct sessions with Amodei.

The regulatory standoff began earlier this month when the US government directed the company to restrict access to its Claude Mythos 5 and Claude Fable 5 models. The directive followed concerns that foreign users, including some employees, may have accessed the systems, alongside reports of potential vulnerabilities. Amazon researchers reportedly alerted federal officials to possible jailbreak techniques (methods used to bypass safety guardrails) shortly after the models were launched. In response to the directives, Anthropic temporarily withdrew access to both models for all users.

Technical meetings are currently underway to determine the evidence required to demonstrate the effectiveness of Anthropic's safety safeguards. While President Donald Trump had previously acknowledged the company's responsible response to the directive, the restrictions remain in effect. Anthropic continues to advocate for maintaining state-level AI protections until a comprehensive federal law is established, citing the importance of independent safety testing. Meanwhile, cybersecurity researchers maintain that AI guardrails remain temporary measures, as dedicated users may eventually discover methods to circumvent security features. Officials and company representatives are continuing dialogue at both technical and senior leadership levels to explore potential conditions for lifting the current access bans.

Negotiations between the White House and Anthropic concerning AI model restrictions appear to be improving, with leadership in the talks shifting away from CEO Dario Amodei. According to a report published on June 25, 2026, discussions are now directed by Anthropic co-founder Tom Brown and head of public policy Sarah Heck. Sources familiar with the process indicated that officials find the new team more approachable and effective at engagement compared to previous direct sessions with Amodei.

The regulatory standoff began earlier this month when the US government directed the company to restrict access to its Claude Mythos 5 and Claude Fable 5 models. The directive followed concerns that foreign users, including some employees, may have accessed the systems, alongside reports of potential vulnerabilities. Amazon researchers reportedly alerted federal officials to possible jailbreak techniques (methods used to bypass safety guardrails) shortly after the models were launched. In response to the directives, Anthropic temporarily withdrew access to both models for all users.

Technical meetings are currently underway to determine the evidence required to demonstrate the effectiveness of Anthropic's safety safeguards. While President Donald Trump had previously acknowledged the company's responsible response to the directive, the restrictions remain in effect. Anthropic continues to advocate for maintaining state-level AI protections until a comprehensive federal law is established, citing the importance of independent safety testing. Meanwhile, cybersecurity researchers maintain that AI guardrails remain temporary measures, as dedicated users may eventually discover methods to circumvent security features. Officials and company representatives are continuing dialogue at both technical and senior leadership levels to explore potential conditions for lifting the current access bans.