Potentially the first time I’ve felt a distinct and persistent sense of moral unease about alignment research.

The panel asks how models infer what graders reward, why good behaviour can come from the wrong reason, and whether that difference can be measured. The conversation moves through promise-breaking, grader awareness, reward hacking, scheming, opaque reasoning and corrigibility, then turns to a detailed walkthrough of the contrastive-belief method and what its results do and do not show. The o3 results discussed here concern an intermediate checkpoint without safety training.

How Researchers Test AI for Hidden Goals — Apollo Research - YouTube ↗