Skip to content

R0043/2026-04-01/Q003

Query: Has the vocabulary gap itself been identified as a problem in the AI safety or AI governance literature? Are there researchers or organizations working to create a shared taxonomy that bridges AI safety terminology with regulated-industry terminology?

BLUF: The broader AI terminology gap is well-recognized, with multiple researchers and organizations proposing solutions. However, the specific sycophancy vocabulary gap — the absence of cross-domain terminology for AI agreement-seeking behavior — has not been prioritized in any identified taxonomy, glossary, or framework. Even the most comprehensive efforts (53-threat taxonomy, 100+ term glossary) exclude sycophancy and all related behavioral terms. The gap persists because sycophancy falls between categories: it is not a governance concept, a security threat, or a process term — it is a behavioral model property that current taxonomy efforts have not yet addressed.

Probability: N/A (open-ended query) | Confidence: Medium-High


Summary

Entity Description
Query Definition Query text, scope, status
Assessment Full analytical product with taxonomy coverage analysis
ACH Matrix Evidence x hypotheses diagnosticity analysis
Self-Audit ROBIS-adapted 5-domain audit

Hypotheses

ID Hypothesis Status
H1 Gap recognized with active taxonomy efforts Supported (at broad level)
H2 Gap not recognized Eliminated
H3 Broader gap recognized but sycophancy excluded from all efforts Supported (best fit)

Searches

ID Target Results Selected
S01 Vocabulary gap recognition and taxonomy efforts 20 4

Sources

Source Description Reliability Relevance
SRC01 Trilateral Research terminology gap analysis Medium-High High
SRC02 CSIRO/UNSW evaluation framework Medium-High Medium-High
SRC03 Huwyler 53-threat taxonomy Medium-High Medium
SRC04 IAPP 100+ term governance glossary High Medium

Key Organizations Working on AI Terminology

Organization Effort Includes Sycophancy?
Trilateral Research AI terminology gap diagnosis + solutions No
CSIRO / UNSW Harmonised evaluation terminology framework No
Huwyler (IE / Capgemini) 53-threat cross-domain taxonomy No
IAPP 100+ term AI governance glossary No
NIST AI RMF + 600-1 GenAI Profile No (uses "confabulation")
Cyber Risk Institute Financial services AI risk framework No
Roytburg & Miller (CMU / Emory) Network analysis of safety-ethics divide Identifies structural barrier to terminology diffusion

Revisit Triggers

  • NIST, ISO, or IEEE publishes a cross-domain taxonomy or glossary that includes sycophancy or an equivalent term
  • A major AI governance conference includes a session on sycophancy terminology harmonization
  • IAPP or similar professional body adds sycophancy-related terms to their glossary
  • Regulatory body adopts "sycophancy" or equivalent in formal guidance
  • The Roytburg & Miller safety-ethics bridge analysis is updated showing reduced homophily