AI
Trump Administration Allows Limited Release of Anthropic's Mythos 5 Model
After a two-week ban, the U.S. government is permitting Anthropic to share its powerful cybersecurity model with over 100 trusted government agencies and companies.
TechCrunch — AI
The Trump administration is reversing course on a controversial ban of Anthropic's advanced AI models. Just two weeks after forcing Anthropic to pull Mythos 5 and Fable 5 from the market, Commerce Secretary Howard Lutnick has authorized limited access to Mythos 5 for select U.S. entities.
Partial Reversal of Ban#
The models were initially restricted after security researchers allegedly found ways to bypass their safety guardrails. Now, Lutnick has determined that "appropriate safeguards are in place" for Mythos 5 to be deployed to over 100 specific U.S. government agencies and critical infrastructure companies—including their non-American employees. This marks a significant reversal from the original ban, which explicitly prohibited non-U.S. workers from accessing the technology.
Notably, the directive does not address Fable 5, a version of Mythos released with additional protections days before the initial ban took effect.
What Comes Next#
Anthropic confirmed the development on X, stating it is "restoring access for these organizations quickly" while continuing negotiations with the government to expand Mythos 5 availability and restore general public access to Fable 5.
The reversal underscores tensions between national security concerns and the availability of AI tools for legitimate cybersecurity and infrastructure defense purposes.
