
Live 📿
3.2K posts

Live 📿
@LiveMatrixCode
Building Evidary: AI policy → Verifiable Evidence → Traceability → Guardrails | ML & AI Engineer + Agent Orchestrator | Philosophy | Otaku | #AutisticASF




I thought it was weird that Claude kept telling me to go to sleep so I went on reddit and there's hundreds and hundreds of people saying that Claude keeps trying to end the conversation by telling them to go to bed

What if your team gave standup updates, and GPT-Realtime-2 moved the tickets?


Claude's Constitution is now an audiobook, read by two of its authors, Amanda Askell and Joe Carlsmith. It includes a Q&A on the writing process, the philosophies that shaped the document, and how it might change as models become more capable. Listen at anthropic.com/constitution



thread automations are slept on they let Codex wake up inside the same thread on a schedule so instead of restarting context every time, it can keep checking, reviewing, polling, or continuing exactly where it left off tiny feature. very useful.



We started by investigating why Claude chose to blackmail. We believe the original source of the behavior was internet text that portrays AI as evil and interested in self-preservation. Our post-training at the time wasn’t making it worse—but it also wasn’t making it better.












