The US administration has imposed export restrictions on Anthropic's artificial intelligence models, Fable 5 and Mythos 5. This decision, it turns out, stems from concerns that a group linked to China may have gained access to the Mythos model. This was the key trigger that prompted the White House to act quickly and decisively.

The China Factor and Other Triggers

According to my information, suspicions regarding an organization with ties to the People's Republic of China have caused serious alarm within the Donald Trump administration. Details — exactly who, how, and when this occurred — remain unclear for now. Concurrently, another important factor was reports of a possible bypass of the Fable 5 model's safeguards. Anthropic itself insists that this involved a narrow, non-universal method that did not grant the model unique capabilities and posed no real threat.

Timeline of Events and Reactions

On June 11, Amazon CEO Andy Jassy notified the administration of the possibility of bypassing Fable 5's restrictions. This occurred as part of a model review. The very next day, US Secretary of Commerce Howard Lutnick sent a letter to Anthropic CEO Dario Amodei demanding compliance with export controls. Officials allegedly attempted to negotiate a voluntary suspension of access, but the startup refused. According to another version, close to the company, the White House gave only 90 minutes to shut down the models without providing clear explanations.

White House advisor David Sacks stated that the vulnerability warning came from a reliable partner trusted by both sides. He also emphasized that "the ball is in Anthropic's court," and that the export controls will be lifted once the issue is resolved.

Anthropic's Position: Misunderstanding or Threat?

Anthropic strongly disagrees with the authorities' actions. The company claims the letter contained no specific details about risks to national security. After conducting its own investigation, the startup concluded that the demonstrated bypass technique only allows finding "a small number of previously known minor vulnerabilities." Moreover, Anthropic states that other publicly available models perform this task just as well. The company emphasizes that it has spent "thousands of hours" testing safeguards jointly with the US and UK governments, and no universal method for mass bypass has been found.

My analysis: This incident is a clear signal that geopolitical tensions are directly encroaching on the field of AI development. The White House, fearing technology leaks, is acting preemptively, possibly even at the expense of innovation. For the market, this means increased uncertainty and volatility, especially for projects related to decentralized AI, as we have already observed with the rise of tokens like Venice and Morpheus. The battle for control over advanced AI models is just beginning, and it will be fierce.