According to the latest report by foreign media outlet Semafor, the White House has decided to restrict exports of Anthropic’s “Mythos” artificial intelligence model, partly due to concerns that it could be accessed by China-linked entities. If the Chinese side were indeed granted access to Mythos 5 or Fable 5, it would be considered a serious national security threat. The report points out that in addition to direct access, the Chinese government may attempt reverse engineering of the model using “distillation” techniques—leveraging more advanced models as “teachers” to train an “student” model with similar behavioral capabilities, thereby replicating its core functions. This pathway has also triggered heightened vigilance among U.S. officials. Notably, the White House has yet to confirm this reporting. In posts on social media platform X, Trump advisor David Sacks did not mention China but instead focused on rumors regarding potential “jailbreak” vulnerabilities in the Fable and Mythos models. However, Anthropic explicitly denied these claims. When approached for comment by Semafor, Anthropic had no formal response at the time, though a spokesperson told the outlet that discussions surrounding export control policies have not mentioned factors related to China. If Mythos was indeed accessed by parties linked to China, this marks another instance where one of Anthropic’s most powerful models suffered security breaches. The company previously stated that Mythos posed too great a risk and power level for public release. Prior reports indicated that members of a Discord group illegally obtained two weeks’ worth of usage rights before Anthropic discovered the vulnerability and revoked access.