US Blocks Anthropic AI Access After Amazon CEO Tip

The U.S. has blocked overseas access to Anthropic's new AI models, Mythos and Fable5. This decision followed Amazon CEO Andy Jassy alerting U.S. officials, including Treasury Secretary Scott Bessent, to potential vulnerabilities. Amazon researchers reportedly found a way to uncover security vulnerability information through specific prompts in Fable5. After security reviews, authorities asked Anthropic to fix the issue or suspend the models. The government then decided to block foreign access, a measure signed by Trump. Anthropic complied, blocking all user access, even for its own foreign national researchers. The company maintains the vulnerability was not a serious threat and similar information is available from other public models. Cybersecurity experts like Andrew Morris agree, noting Fable5 could identify bugs but not generate exploit code. This action comes as Anthropic prepares for an initial public offering, and some analysts believe the move could impact its valuation and benefit competitors.