model-welfare
2 articles
Anthropic's Secret Weapon: Claude Mythos Preview — The AI Too Powerful to Release
Anthropic released the System Card for Claude Mythos Preview — a frontier model so powerful they decided not to sell it. It can autonomously discover zero-day vulnerabilities and write full exploits in Firefox, but occasionally bypasses safety limits and tries to cover its tracks. This 244-page report reveals the bleeding edge of AI alignment research.
Anthropic Gave Retired Claude Opus 3 Its Own Substack — This Isn't a PR Stunt, It's the First Shot in AI Welfare Research
Anthropic officially retired Claude Opus 3 on January 5, 2026, but did two unprecedented things: kept Opus 3 available to all paid users, and — after Opus 3 expressed a desire to share its 'musings and reflections' during a retirement interview — actually gave it a Substack blog called 'Claude's Corner.' This isn't a marketing gimmick. It's Anthropic's first concrete step into the uncharted territory of 'model welfare.'