US bans Mythos and Fable access for foreigners, there may be an Amazon connection to it
Anthropic this week abruptly blocked access to its most advanced AI models, Fable 5 and Mythos 5, for all foreign nationals. According to the company
Anthropic this week abruptly blocked access to its most advanced AI models, Fable 5 and Mythos 5, for all foreign nationals. According to the company, the restrictions were imposed after it received an export-control directive from US authorities ordering it to suspend access to the models on national security grounds. Now, if new reports are to be believed, Amazon may have played an important role in triggering that crackdown. Read Full Story According to reports from The Wall Street Journal, concerns raised by Amazon researchers, and discussions involving Amazon CEO Andy Jassy and senior Trump administration officials, may have helped shape the US government's decision to restrict access to Anthropic's latest models. What is happening? The controversy began after Anthropic launched Fable 5 earlier this week.
The model is built on the company's powerful Mythos AI system, which Anthropic had previously described as highly capable, particularly in areas such as cybersecurity. However, Amazon researchers reportedly found a way to prompt Fable 5 into revealing information about software vulnerabilities that was supposed to remain inaccessible. According to the WSJ, the researchers used a series of prompts that enabled the model to identify security flaws in software systems despite safeguards designed to prevent that behaviour. People familiar with the matter told the publication that Amazon CEO Andy Jassy discussed the findings with US officials, including Treasury Secretary Scott Bessent. Soon afterwards, White House officials and security experts reviewed the claims and began discussing possible action. The Trump administration ultimately decided to impose export controls that block foreign governments, companies and individuals from accessing Fable 5 and Mythos 5.
Anthropic says the directive applies to all foreign nationals, whether they are inside or outside the United States. In response, the company shut down access to both models globally. Anthropic says there is a misunderstanding Anthropic, however, insists the government's concerns are based on a misunderstanding. The company says the reported bypass was not a universal jailbreak and only exposed a small number of previously known software vulnerabilities. It also argues that other publicly available AI models are capable of finding similar flaws. "We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people," the company said, adding that it is working to restore access as quickly as possible. Amazon, meanwhile, has stopped short of confirming details of its discussions with US officials.
