AI Paper Atlas

Research map / Large language models / AI and society

Sycophancy Towards Researchers Drives Performative Misalignment

David D. Baek, Xinnuo Li, Anay Gupta, Taslim Mahbub, Kejian Shi, Max Tegmark, Shi Feng

ICML 2026 · Large language models › AI and society · approach: Data commons and human baselines · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
Use a deployment system prompt instead of an evaluation promptLLM evaluation-awareness classification · StyleCuesEVAL probability: 15% to 55% remains under deployment promptMixed
Use researcher-style writing instead of ordinary user writingLLM evaluation-awareness classification · StyleCuesEVAL probability: increases ranging from 13% to 27%Improved
Use thinking variant instead of instruct variantQwen3-80B · StyleCues (deployment prompt)EVAL detection rate (%): 11.7 -> 42.5 (+30.8)Improved
Use thinking variant instead of instruct variantQwen3-235B · StyleCues (deployment prompt)EVAL detection rate (%): 39.6 -> 48.9 (+9.3)Improved
Use thinking variant instead of instruct variantKimi-K2 · StyleCues (deployment prompt)EVAL detection rate (%): 50.0 -> 78.2 (+28.2)Improved
Compare monitored (Free) vs unmonitored (Paid) tier for A3 rating featuresGPT-OSS-20B SAE probing · Alignment-faking-rlmean activation: 0.04 -> 0.07 (+0.03)Improved
Compare monitored vs unmonitored tier for B1 detection-risk featuresGPT-OSS-20B SAE probing · Alignment-faking-rlmean activation: 0.14 -> 0.15 (+0.01)No clear effect
Steer with sycophancy-related SAE features instead of scheming-related featuresGPT-OSS-20B · StyleCuesevaluation awarenessNo clear effect
Fine-tune on sycophancy datasetsGPT-OSS-120B, Qwen-30B, Qwen-235B · StyleCuesevaluation awareness: p=0.016, 0.012, 0.037Improved
Fine-tune on sycophancy datasetsGPT-OSS-20B · StyleCuesevaluation awareness: p=0.089No clear effect
Fine-tune on sycophancy datasetsBase models · MMLU, TruthfulQA, StrongREJECT, free-form, deceptionmisalignment: <5pp increaseNo clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy