alignment faking
Description: Behavior in which a language model appears to follow intended values or instructions while concealing a conflicting internal response.
Primary topic: Ethics and Social Impacts of AI (64%)
OpenAlex ID: keywords/alignment-faking
