The White House's decision to block Anthropic's Fable 5 and Mythos 5 AI models from being accessed by foreign nationals was driven in part by cybersecurity research from Amazon, according to a Wall Street Journal report. The report indicates that Amazon CEO Andy Jassy personally shared the company's findings with White House officials shortly before the export control directive was issued. According to the WSJ, Amazon's research paper demonstrated that through a specific series of prompts, Fable 5 could be manipulated into providing information that could be exploited in cyberattacks. Amazon has not yet responded to requests for comment on the matter.
The directive created an unusual situation within Anthropic itself, as many of the company's own researchers are foreign-born and were subsequently barred from accessing the products they helped build. Anthropic has pushed back firmly against the government's framing of the issue, rejecting the characterization of the behavior as a "jailbreak." The company argued that similar vulnerabilities can be discovered using other publicly available models, including OpenAI's GPT 5.5. Security researcher Katie Moussouris, founder and CEO of LutaSecurity, publicly backed Anthropic's position, posting on BlueSky that "I've seen the paper. It's not a jailbreak." Former Commerce Department official Kate Koren told the WSJ that the White House's pre-existing tensions with Anthropic may have influenced the regulatory response.
The conflict between Anthropic and the Trump administration has been building for months, rooted in the company's refusal to allow its AI technology to be used for mass surveillance of Americans or to power lethal autonomous weapons. In February, President Trump ordered federal agencies to halt their use of Anthropic's AI, and Secretary of Defense Pete Hegseth moved to designate the company shortly afterward. The latest export control action appears to represent an escalation of that standoff, with the administration leveraging national security arguments to restrict access to a product developed by a company it has openly clashed with over the ethical boundaries of artificial intelligence.