White House Export Limits on Anthropic's AI Over China Concerns

The White House imposed export controls on Anthropic's Mythos AI model due to suspicions a China-linked group had accessed it. The Trump administration directed Anthropic to limit access to Mythos and its consumer version, Fable 5, to US citizens. This meant even foreign national employees of Anthropic would be prohibited from using the models. Anthropic responded by removing the models from the market completely. Mythos, launched in April, was designed to find bugs in computer code, a capability Anthropic says poses a danger if exploited by malicious actors. An adviser to President Trump, David Sacks, stated that the government was warned Fable 5 could be "jailbroken," meaning its safety guardrails could be removed. Sacks alleged that Anthropic's CEO, Dario Amodei, dismissed this as not a serious risk. A person close to the White House indicated Amazon informed the government about the jailbreak. An Anthropic spokesperson said the White House did not raise Chinese access concerns in their discussions about the jailbreak and export controls. This situation highlights the complex national security implications of powerful AI models.