DeepMind’s Take on How To Create a Benign AI ChatGPT Hacks BLACK AI PRO January 19, 2025 · 0 Comment The paper “Scalable agent alignment via reward modeling: a research direction” is available here: 1. source deepmind