Skip to content

R0043/2026-04-01/Q003/SRC04/E01

Research R0043 — Sycophancy Vocabulary
Run 2026-04-01
Query Q003
Source SRC04
Evidence SRC04-E01
Type Factual

IAPP governance glossary covers 100+ terms but excludes sycophancy and related concepts

URL: https://www.aigl.blog/key-terms-for-ai-governance-2/

Extract

The IAPP glossary spans 100+ terms including: - Governance concepts: accountability, transparency, fairness, contestability - Technical AI design: neural networks, transformer models, RLHF - Risk terminology: adversarial attacks, bias, robustness - Data lifecycle terms

Terms NOT included: sycophancy, agreeableness bias, acquiescence bias, automation bias, overreliance, automation complacency.

The glossary includes "hallucinations" as part of "Risk and safety terminology" but does not define or mention agreement-seeking behavior.

Commentary notes that "Terms like fairness or explainability are defined descriptively but without detailing trade-offs between different fairness metrics" and that terms "central to current AI safety discourse are absent."

JUDGMENT: The IAPP glossary's omission of sycophancy and ALL related behavioral safety terms — despite covering 100+ AI governance concepts — demonstrates that sycophancy is not yet part of the AI governance vocabulary. This glossary represents the mainstream governance community's understanding of what terms need defining; sycophancy is not yet on their radar.

Relevance to Hypotheses

Hypothesis Relationship Strength
H1 N/A The glossary effort exists but does not address sycophancy
H2 Supports At the sycophancy-specific level, the gap is not recognized
H3 Strongly supports Broader terminology efforts exist but exclude sycophancy

Context

The IAPP is the largest global privacy and AI governance professional organization. Its glossary represents the practical vocabulary that governance professionals use. The absence of sycophancy from this vocabulary is significant for the vocabulary gap question.