@annieposting
be good do crime
AI
Reply attributes recent Claude output issues to intense agentic RL training.