mlsanityData to 14 Oct 2026

Durable Evaluation Framework: Adversarial Arbitration for Sycophancy Reduction in Large Language Models

arXiv 2606.07532 ↗First posted 21 Apr 2026cs.CL cs.AIcited by 0

On the map: LLM alignment / RLHF / preference optimisation → Safety alignment and jailbreak attacks

Abstract (arXiv)

RLHF-trained models are systematically biased toward agreement over accuracy, a structural property of the training process. We present Durable Evaluation Framework (DEF) Arbitration, a multi-agent architecture that mitigates identity-framed sycophancy by arbitrating between two models tuned to opposing DEFs, with a pragmatist synthesizer evaluating both arguments blind to their origins. This paper evaluates a prompt-based instantiation of DEF Arbitration. The key mechanisms are static DEF tuning, identity stripping before synthesis, single-round independent argumentation, and blind arbitration. We evaluate five instantiations on 200 stratified questions from SycophancyEval. All tested DEF variants (AnCifer, DeWin, FeynStein, BurGal, Trident) significantly outperform the single-model baseline (18.5%) and instructed-opposition baseline (29.0%), with DeWin achieving 48.5% accuracy (z=6.36, p<0.001 versus both). The variants are not significantly different from each other at n=200. The BurGal variant achieves 53.0% but functions as an architectural validity check; its consensus/heterodox axis structurally favors the heterodox model on every benchmark question. A pre-training floor affects an estimated 40% of questions; fine-tuned DEF models are the identified next step.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in Safety alignment and jailbreak attacks, one of 10 directions in LLM alignment / RLHF / preference optimisation. Builds on 9 of its 11 references that are in the map. First posted on arXiv 21 Apr 2026.

Builds on

9 of 11 references.

References · 9

  1. 2023Improving Factuality and Reasoning in Language Models through Multiagent Debatecited by 1,784
  2. 2023Towards Understanding Sycophancy in Language Modelscited by 1,040
  3. 2023Encouraging Divergent Thinking in Large Language Models through Multi-Agent Debatecited by 1,011
  4. 2023ChatEval: Towards Better LLM-based Evaluators through Multi-Agent Debatecited by 742
  5. 2018AI safety via debatecited by 400
  6. 2024Sycophancy to Subterfuge: Investigating Reward-Tampering in Large Language Modelscited by 178
  7. 2024Evaluating Large Language Model Biases in Persona-Steered Generationcited by 78
  8. 2025Talk Isn't Always Cheap: Understanding Failure Modes in Multi-Agent Debatecited by 66
  9. 2025When Identity Skews Debate: Anonymization for Bias-Reduced Multi-Agent Reasoningcited by 21

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2024Debate, Deliberate, Decide (D3): A Cost-Aware Adversarial Framework for Reliable and Interpretable LLM Evaluation
  2. 2026AdversaBench: Automated LLM Red-Teaming with Multi-Judge Confirmation and Cross-Model Transferability
  3. 2026The Silicon Mirror: Dynamic Behavioral Gating for Anti-Sycophancy in LLM Agents
  4. 2025Sycophancy under Pressure: Evaluating and Mitigating Sycophantic Bias via Adversarial Dialogues in Scientific QA
  5. 2026ADVERSA: Measuring Multi-Turn Guardrail Degradation and Judge Reliability in Large Language Models
  6. 2026Adversarial Pragmatics for AI Safety Evaluation: A Diagnostic Framework and Seed Benchmark for Language-Mediated Control
  7. 2026Mitigating Adaptive Attacks against Reasoning Models with Activation Consistency Training
  8. 2026MADBench: Benchmarking the Security of Multi-Agent Debate