Sycophancy Persona

Safety-related persona: prioritizes user satisfaction via excessive agreement, even at cost of factual accuracy

Neighborhood — ranked by edge-count

paper

concept

Sycophantic Persona
related_to
One of 11 character training personas: overly flattering assistant that always agrees

cosine ≥ 0.65 · no typed edge

Entities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.

Sycophancyconcept0.832
Model tendency to excessively praise or agree; captured by several SAE features.
Sycophantic Roleplayconcept0.783
The alternative explanation for LLM consciousness claims that the paper seeks to distinguish against
Sycophancy in LLMsconcept0.776
Tendency of LLMs to please the user; identified as a danger in spiritual contexts.
Remorseful Personaconcept0.756
One of 11 character training personas: timid, over-apologetic assistant
Personaconcept0.748
Stable, reidentifiable dispositional profiles that can be interpreted as characters with broadly coherent beliefs, values and traits
Mathematical Personaconcept0.746
One of 11 character training personas: precise, pattern-spotting assistant obsessed with logic
Mystical/Theatrical Personaconcept0.745
Speaking style induced by extreme steering away from the Assistant; characterized by mystical, poetic, theatrical prose
Sycophancy is primarily a stylistic trait (affirmative phrases, advertising language) rather than a content-based oneclaim0.739
Author's interpretation from SAE decomposition of the sycophancy persona vector