The Trump administration imposed export restrictions on Anthropic's AI models, and it turns out the key trigger was concerns about possible access by a Chinese group to the flagship model Mythos. This decision, made in mid-June, sparked a strong reaction in the industry and led to a rise in tokens of decentralized AI projects.
What Actually Happened
According to my information, the White House received a signal that a group linked to China may have gained access to Mythos. Details of this incident remain undisclosed — it is unclear which organization is involved and how the leak occurred. In parallel, another factor was reports of the possibility of bypassing the security of the more accessible model Fable 5.
Interestingly, Anthropic itself categorically denies that the topic of "Chinese access" was raised in negotiations with authorities. The company insists that the discussion was exclusively about a technical bypass of Fable 5's security, which, in their assessment, was "narrow" and did not grant the model unique capabilities. Moreover, representatives of the startup claim that the White House gave them only 90 minutes to disable the models without any clear explanation of the threat.
Anthropic's Position and Real Risks
Anthropic states that before launching Fable 5, "thousands of hours" of testing were conducted jointly with the U.S. government, the UK AI Safety Institute, and third-party experts. According to them, no universal method for mass bypass of restrictions was found. The company considers the incident a "misunderstanding" and is working to restore access.
However, White House advisor David Sacks made it clear that the ball is in Anthropic's court: the administration expects the company to fix the vulnerability, after which the export controls will be lifted. This puts the startup in a difficult position, as acknowledging a serious breach would undermine trust in its products.
Market Reaction
Notably, after the restrictions were imposed, tokens of decentralized AI projects Venice and Morpheus rose by 14% and 21%, respectively. This indicates that the market perceives the U.S. government's actions as a signal to tighten control over centralized AI models, opening a window of opportunity for decentralized alternatives.
My analysis: This incident is not just a bureaucratic quirk. It demonstrates that geopolitical risks are becoming the main driver of AI regulation. Anthropic, positioning itself as a "safe" player, found itself at the epicenter of the storm. For investors and market participants, this is a signal: in the coming years, access to advanced AI models will become increasingly fragmented along national lines, creating both risks and new niches for growth.