Training large language models on narrow tasks can lead to broad misalignment #1 Post by thebeardisred » Fri, Jan 16, 2026, 6:17 AM UTC Training large language models on narrow tasks can lead to broad misalignmentnature.com
Re: Training large language models on narrow tasks can lead to broad misalignment #2 Post by merksittich » Fri, Jan 16, 2026, 8:14 AM UTC Discussions on the manuscript/preprint:https://news.ycombinator.com/item?id=43176553https://news.ycombinator.com/item?id=44554865