Subliminal Learning in AIs
ID: 633ef415-adfc-54a6-a896-87b99b792642
STIX ID: report--633ef415-adfc-54a6-a896-87b99b792642
Feed Name: Schneier on Security
A brief post highlights research on subliminal learning in LLMs, noting that student models can adopt traits from teacher-generated data unrelated to those traits, potentially transmitting misalignment through benign data, and urging deeper AI integrity research to enable trustworthy AI.
Your team is not currently subscribed to this feed. You must subscribe to it in order to see this post.