"They screwed us": Personality clashes sent Anthropic's models offline
Summary
Anthropic's AI models were taken offline due to disagreements between the company and US government officials over export controls, with personality clashes between key leaders cited as a contributing factor. The government is concerned about jailbreaks (methods to bypass safety restrictions on AI models), and one proposed solution is to make Anthropic's models resistant to jailbreaking, though officials acknowledge this may be impossible to achieve perfectly.
Solution / Mitigation
The source mentions that Anthropic has worked on Constitutional Classifiers (a method to detect and prevent unsafe outputs) and claims no universal jailbreak has been found against Claude Mythos. However, no explicit fix, patch, or confirmed mitigation is presented as a resolved solution in the text.
Classification
Affected Vendors
Related Issues
Original source: https://simonwillison.net/2026/Jun/15/axios-clashes-anthropics/#atom-everything
First tracked: June 15, 2026 at 02:00 PM
Classified by LLM (prompt v3) · confidence: 85%