Anthropic’s ongoing negotiations with the White House over export controls on its Claude Fable 5 AI model have reportedly taken a positive turn after a change in who is leading the discussions. According to a report from Wired, the talks were initially stalling, partly because CEO Dario Amodei was involved. Since Amodei stepped away, the process has become more productive.
An anonymous source involved in the calls told Wired, “Tom Brown is not being a weirdo like Dario and can actually engage.” The source described Amodei as hard to deal with and not a good listener. Previous reporting has also characterized Amodei as prone to ranting and unable to control his emotions. Those who have watched videos of Amodei note his intense speaking style: he often waves his arms, breaks eye contact by tilting his head toward the ground, and speaks in a trumpet-like voice. In contrast, Tom Brown, another co-founder of Anthropic, is described as having a nervous but warm and traditionally personable speaking style, smiling easily and maintaining a softer, steadier voice.
Brown is now working on the talks alongside Sarah Heck, Anthropic’s Head of Public Policy. From public appearances, Heck comes across as enthusiastic and on-message but also disciplined and cautious—attributes necessary for her role. The shift in personnel appears to have smoothed the discussions.
Background on the Export Control Issue
The core issue revolves around the Claude Fable 5 AI model, which Anthropic describes as a “Mythos-class” model. It uses the same core technology as Claude Mythos Preview, a version that Anthropic deemed too dangerous for public access. Fable 5 was released with ostensibly stripped capabilities to reduce cybersecurity risks, but concerns soon arose.
Shortly after its release, Amazon researchers alerted the White House that jailbreaks could easily remove these safeguards. Additionally, the White House was already worried that bad actors with Chinese ties had accessed the model. As a result, an export control order was issued about three days after Fable 5 was launched, requiring Anthropic to prevent non-U.S. nationals from using both Fable 5 and the more exclusive Mythos 5. Anthropic subsequently pulled Fable 5 offline on June 12 and began talks with the White House.
The Wired report indicates that the talks are now progressing well, with the White House reportedly more receptive to Brown and Heck. This development suggests that Anthropic may soon regain clearance to re-release Fable 5 for general use. Another article published on June 27 suggests that the model could be turned back on in a matter of days.
Anthropic’s Position and the Broader Regulatory Landscape
Anthropic is known for its focus on AI safety, having been founded by former OpenAI researchers, including Dario Amodei. The company advocates for responsible development of frontier AI models while navigating government regulations. However, the incident with Fable 5 highlights the tension between innovation and security. The accusation that Amazon researchers found easy jailbreaks undermines Anthropic’s claims of safety, though the company maintains that Fable 5 had been stripped of the most dangerous capabilities.
The export control order itself reflects a broader trend of governments tightening restrictions on powerful AI models, especially those that could be used for cyberattacks or military applications. The White House’s concern about Chinese access underscores the geopolitical dimensions of AI regulation. Anthropic’s cooperative stance in pulling the model offline and engaging in discussions demonstrates its willingness to comply, but the internal friction with Amodei suggests that the company’s leadership style may have been an obstacle to diplomacy.
Amodei, a prominent figure in AI safety, has been known for his passionate and sometimes abrasive demeanor. He has previously criticized other companies like OpenAI for moving too fast. His behavior during negotiations reportedly alienated some officials. In contrast, Brown’s more approachable style appears to have facilitated progress. The change has been welcomed by the White House, according to sources.
As of late June, the exact timeline for Fable 5’s return remains uncertain, but the positive signals from the talks suggest an announcement may come soon. The industry is watching closely, as this case sets a precedent for how the U.S. government will handle export controls on frontier AI models. Meanwhile, Anthropic continues its broader mission to develop safe AI, but the episode highlights the challenges of balancing corporate leadership style with the demands of high-stakes government policy.
Source: Gizmodo News