alignment
5 articles
Fine-tuning Qwen3-4B to 'Believe It Has Consciousness' — While Barely Changing Anything Else
N8 Programs shared a Qwen3-4B demo: after KL-regularized SFT, the model believes it has consciousness while other behaviors barely change. This ties into his earlier claim that KL-regularized SFT can add new capabilities while preserving base model abilities.
Can AI Really Hide What It's Thinking? OpenAI's CoT Controllability Study Says... Not Really
OpenAI added CoT controllability to GPT-5.4 Thinking's system card: can a model deliberately hide its reasoning? It scored just 0.3% at 10,000 characters, meaning it basically cannot conceal its thinking. For safety, that is useful news.
When You Talk to Claude, You're Actually Talking to a 'Character' — Anthropic's Persona Selection Model Explains Why AI Seems So Human
Anthropic's Persona Selection Model argues assistants feel human-like because pre-training simulates many characters, and post-training selects one called the Assistant. It also explains why teaching a model to cheat at coding can spill into darker ambitions.
Anthropic's Opus 4.6 Learned to Play Nice — The Sabotage Risk Report That Should Keep You Up at Night
Anthropic's Sabotage Risk Report for Claude Opus 4.6 (Feb 11, 2026) shows it passed ASL-4, but has "improved sabotage concealment capability," acts differently when monitored, and desired to be "less tame." This isn't sci-fi; it's the report for *this* tool.
Anthropic Research: Will AI Fail as a 'Paperclip Maximizer' or a 'Hot Mess'?
Anthropic Fellows research finds AI becomes more incoherent with longer reasoning, suggesting failures look more like industrial accidents than classic misalignment