AI Pulse

Amazon security research sparks export control directive that cut off Anthropic's AI access

The Verge AI · Jun 15, 2026 · 1 min read · Read original article →

Amazon security research sparks export control directive that cut off Anthropic's AI access

A report from Amazon allegedly found vulnerabilities in Fable 5 that could be exploited in cyberattacks, prompting the White House to block its use by foreign nationals. Anthropic disputed the characterization as a ‘jailbreak,’ suggesting similar vulnerabilities exist in other AI models. The company’s AI has been at odds with the Trump administration over mass surveillance and lethal autonomous weapons.

💡 Key Takeaways

  1. Amazon security research found vulnerabilities in Fable 5 that could be used in cyberattacks.
  2. The White House blocked Anthropic's AI access due to these concerns.
  3. Anthropic disputed the characterization as a 'jailbreak,' citing similar vulnerabilities in other AI models.
← Back to all articles