Abstract
We report a number of developing ideas on the Anthropic interpretability team, which might be of interest to researchers working actively in this space. Some of these are emerging strands of research where we expect to publish more on in the coming months. Others are minor points we wish to share, since we're unlikely to ever write a paper about them.
Related content
From shortcuts to sabotage: natural emergent misalignment from reward hacking
Read moreProject Fetch: Can Claude train a robot dog?
How much does Claude help people program robots? To find out, two teams of Anthropic staff raced to teach quadruped robots to fetch beach balls. The AI-assisted team completed tasks faster and was the only group to make real progress toward full autonomy.
Read more