White House Blocks Anthropic's Latest AI Models Over Security Concerns After Amazon Research

Summary

The White House has blocked access to Anthropic's Fable 5 and Mythos 5 models following cybersecurity research from Amazon and conversations between Amazon CEO Andy Jassy and government officials. According to the Wall Street Journal, Amazon's research claimed the models could be manipulated through specific prompts to provide information useful for cyberattacks. The export control directive took effect immediately, preventing foreign nationals from accessing the models—a decision that ironically bars many of Anthropic's own researchers, who are foreign-born.

Anthropic disputed the government's characterization of the issue as a "jailbreak," arguing that similar vulnerabilities exist in other publicly available models like GPT 5.5. Some security researchers, including LutaSecurity CEO Katie Moussouris, have sided with Anthropic's assessment. Former Commerce Department official Kate Koren suggested to the WSJ that political considerations may have influenced the decision.

The ban marks an escalation in tensions between Anthropic and the Trump administration, which has repeatedly clashed with the company over its refusal to support mass surveillance or autonomous weapons. In February, Trump instructed federal agencies to stop using Anthropic's AI, and Defense Secretary Pete Hegseth designated the company a supply chain risk. Though the relationship briefly improved as the two worked together on Mythos expansion, the latest action suggests further conflict ahead.

• The ban reflects ongoing tension between Anthropic and the Trump administration over the company's stance on surveillance and autonomous weapons
• Some security experts question whether political motivations influenced the decision beyond legitimate security concerns