Anthropic's negotiations in Washington to lift the ban were unsuccessful; Claude Fable 5 remains globally banned.

robot
Abstract generation in progress
On June 16, U.S. Department of Commerce and Anthropic held an emergency working meeting on the recent AI model export controls implemented on June 15, but the talks ultimately ended without resolution, and Claude Fable 5 remains globally banned. The earlier U.S. government restrictions were mainly driven by security concerns, fearing that once Fable 5's defenses are bypassed through jailbreaking, it could degrade and release powerful cyberattack and vulnerability exploitation capabilities at the Mythos level, which could then be misused by military intelligence agencies of other countries.
This closed-door meeting in Washington was led by Anthropic co-founder and Chief Computing Officer Tom Brown, former Red Team lead Logan Graham, and senior security researcher Nicholas Carlini. Officials from the National Cybersecurity Directorate and the Department of Commerce attended to listen to the discussion. The team reiterated that the U.S. government's concerns about jailbreaking risks for Fable 5 are exaggerated, but officials from the Department of Commerce and researchers from the Center for AI Safety and Innovation (CAISI) were not convinced. The government insisted that model security measures could be bypassed and demanded that Anthropic must thoroughly address jailbreaking vulnerabilities.
Currently, both sides are urgently discussing the next steps. U.S. Secretary of Commerce Howard Lutnick is participating remotely from the G7 summit in Évian-les-Bains, France, and maintains regular calls with Anthropic CEO Dario Amodei. Since both are attending the conference in France, Lutnick and Amodei are expected to hold a face-to-face meeting at the G7 summit site this week. Regarding the ban, over 80 cybersecurity professionals, including executives from Nvidia, Adobe, Zoom, and academic experts, signed an open letter on June 14 to support Anthropic, arguing that the ban not only harms cybersecurity defenders but also weakens the U.S. leadership in AI.
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • Comment
  • Repost
  • Share
Comment
Add a comment
Add a comment
No comments
  • Pinned