The Information Machine

"Dangerous" AI models are coming no matter what

Ars Technica AI · Lily Hay Newman, WIRED.com · 2026-06-17

The US government issued an export-control directive requiring Anthropic to take Claude Fable 5 and Mythos 5 offline for foreign nationals, triggering White House negotiations over how to handle frontier models with acknowledged dual-use cybersecurity and biology capabilities.

Open original ↗

Appears in

Extraction

Topics: ai-regulationexport-controlsdual-use-aiai-safetycybersecurity

Claims

  • The US government issued an export-control directive barring any foreign national from using Anthropic's Fable 5 and Mythos 5 models, forcing Anthropic to take them offline.
  • Anthropic publicly acknowledged that Mythos 5 has advanced dual-use capabilities for both finding and exploiting software vulnerabilities.
  • Anthropic pre-released Mythos 5 privately to a restricted consortium called Project Glasswing before its public launch.
  • Claude Fable 5 was released to the general public with specific capability blocks on biology and cybersecurity responses.
  • Anthropic had been in talks with the White House since Friday but had not yet secured an agreement to reinstate the models at time of publication.

Key quotes

A great deal of advanced usage of AI models is dual use: the same queries that are beneficial in the hands of cybersecurity professionals and biology researchers could be dangerous if available to malicious actors.