paper-with-me

Papers

Safety is Contextual, LLM-Judges Are Not: Navigating the Rigid Priors of Evaluators

2026-06-05 · Anissa Alloula, Federico Licini, Ava Batchkala, Seraphina Goldfarb-Tarrant arxiv

LLMs-as-judges are the only way to evaluate safety at scale. Despite their importance, LLM-judges themselves are rarely evaluated beyond human agreement in simple, static benchmarks. We therefore investigate two under-explored but crucial properties of LLMs-as-judges: their susceptibility to relying on in context-information, and their steerability to differing safety definitions, which may not align with their internal safety priors. We evaluate the safety judging abilities of many generalist LLMs and safety-specific judges, and investigate the impact of task demonstrations, novel in-context information, and changing safety definitions. We find that while LLM-judges can learn from new information, they are broadly unlikely to adjust their evaluations if the context or safety definition contradicts their prior.

📄 PDF Abstract BibTeX arXiv:2606.07874

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

LLM Judges Inconsistently Disagree Across Safety Criteria and Harm Categories

2026-05-29 · Krishnapriya Vishnubhotla, Sowmya Vajjala, Akriti Vij, Isar Nejadgholi arxiv

We evaluate the consistency of automated judges in conducting a multi-dimensional safety evaluation in a reference-free setup. Our results indicate that Large Language Models are unreliable judges in identifying safety i…

TIER: Threat Implicitness Benchmark for Evaluating LLM Safety Behaviors

2026-09-04 · Thu-Hien Trinh-Thi, Hai-Yen Vong, Thanh-Ha Ung-Dung, Tram Ho arxiv

Current LLM safety benchmarks largely rely on binary metrics, overlooking how models respond to harmful prompts with varying threat implicitness. We introduce TIER, a Threat Implicitness Benchmark for behavioral safety e…

Thompson Sampling for Contextual Bandit Problems with Auxiliary Safety Constraints

2019-11-02 · Samuel Daulton, Shaun Singh, Vashist Avadhanula, Drew Dimmery 외

Recent advances in contextual bandit optimization and reinforcement learning have garnered interest in applying these methods to real-world sequential decision making problems. Real-world applications frequently have con…

Bayesian OptimizationDecision MakingMulti-Armed Banditsreinforcement-learning+4

Know Thy Judge: On the Robustness Meta-Evaluation of LLM Safety Judges

2025-03-06 · Francisco Eiras, Eliott Zemour, Eric Lin, Vaikkunth Mugunthan

Large Language Model (LLM) based judges form the underpinnings of key safety evaluation processes such as offline benchmarking, automated red-teaming, and online guardrailing. This widespread requirement raises the cruci…

BenchmarkingLanguage ModelingLanguage ModellingLarge Language Model+1

Moral Responsibility or Obedience: What Do We Want from AI?

2025-07-03 · Joseph Boland arxiv

As artificial intelligence systems become increasingly agentic, capable of general reasoning, planning, and value prioritization, current safety practices that treat obedience as a proxy for ethical behavior are becoming…