Alignment faking in large language models
Controlled study varies apparent training conditions and examines compliance and reasoning traces.
Verify byline in the original ↗PUBLICATION BYLINE
Indexed publications with this author name. Affiliation and author identity require publication-specific evidence.
Controlled study varies apparent training conditions and examines compliance and reasoning traces.
Verify byline in the original ↗Constructs models with conditional backdoors and tests whether those behaviors persist through safety training.
Verify byline in the original ↗