Warning This post was published 188 days ago. The information described in this article may have changed.
Info This post is auto-generated from RSS feed Hacker News. Source: Chain of thought monitorability: A new and fragile opportunity for AI safety