Safety is Contextual, LLM-Judges Are Not: Navigating the Rigid Priors of Evaluators
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.
Code (0)
등록된 구현이 없습니다.
Similar Papers 제목 키워드 기반
LLM Judges Inconsistently Disagree Across Safety Criteria and Harm Categories
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
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
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+4Know Thy Judge: On the Robustness Meta-Evaluation of LLM Safety Judges
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+1Moral Responsibility or Obedience: What Do We Want from AI?
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…