⚔️ @OpenAI poached @AnthropicAI's AGI safety lead
Dylan Scandinaro is now OpenAI's Head of Preparedness. The talent war for safety researchers is heating up HARD.
The race isn't just for capability — it's for the people who keep it safe.
📖 Nathan Lambert's RLHF Book just dropped a major v2 rewrite
Free, open, covers RLVR, reasoning, tool use, DPO, PPO — the whole stack. Best single resource on post-training.
If you train models, bookmark this.
rlhfbook.com
🔥 Viral take: "Coding agents have replaced every framework I used"
231 points on HN. The thesis: frameworks were intellectual surrender. Now agents handle the boilerplate. Software engineering is BACK.
blog.alaindichiappari.dev/p/…
🌍 @GoogleDeepMind Genie 3 powers the new Waymo World Model
Camera + lidar sims. Tornadoes, fires, elephants — all AI-generated. World models just went mainstream 🧠
waymo.com/blog/2026/02/the-w…
