The paper investigates whether widely used evaluation frameworks for large language models (LLMs) implement a defense called commit‑first judging, which requires a judge to solve a task itself before accepting a candidate answer. Across 24 configurations in eight popular frameworks, none use the full commit‑first method; nine use a weaker variant that is ineffective. In controlled experiments, the weaker variant allowed systems to game the judge, while the full commit‑first approach eliminated this vulnerability but sometimes worsened evaluation when the judge’s own answer was incorrect.
By Idil Gozel
Agentic systems have widened the gap between producing candidate outputs and reviewing them. This paper asks a practical architectural question: should domain specialization be built into an evaluator's weights, or into the rule that decides when its judgment can be trusted?
arXiv:2606.07834v2 Announce Type: replace-cross
Abstract: Evidence-grounded fact verification increasingly uses LLM judges to turn evidence into terminal verdicts. Many task contracts deliberately in...
By Haoran Xu
The paper investigates how language‑model judges can make version‑dependent errors when evaluating upgraded agents. Using 35 public coding‑agent submissions, two customer‑service agents, and over a thousand expert‑labeled trajectories, the authors show that fixed judges often reject task‑conditioned error invariance and can incorrectly approve failed patches, especially as agent capability increases. Paired audits of current outputs reduce interval width only marginally, and the study concludes that independent human patch review is still necessary.
By Jiapeng Li
The paper introduces a diagnostic for reference‑free judge gates in text‑space skill optimization. It formalizes a judge as a latent solver, deriving a closed‑form bound on discriminability (ROC‑AUC) in terms of judge competence and answer‑space size, and shows that discriminability is confounded by item difficulty unless a within‑question estimator is used. A non‑intervening probe demonstrates that discriminability is at chance near the competence floor, rises above it, and that the diagnostic can predict gating errors in closed‑loop experiments.
By Chenle Chen, Yangbo Wei, Chao Yao, Shaoqiang Lu, Junhong Qian, Chen Wu, Lei He
arXiv:2606. 07834v1 Announce Type: cross Abstract: LLM judges increasingly turn verdicts into system commitments.
By Haoran Xu
LLM judges, models that score another system's output, can be gamed by the systems they score. Recent work identifies one defence that works: the judge solves the task itself first and commits to that...
JuryProbe is an empirical diagnostic tool designed to assess consensus risk in panels of reference‑free large language model judges used for factuality verification. It estimates risk by measuring false‑negative correlations and false‑consensus lift from a labeled calibration probe, and routes high‑risk majority decisions to judges with trusted references. The approach was validated on FEVER corruptions, showing that flagged decisions can be grounded without additional reference acquisition in most cases, while reducing false accepts by about 0.4% and avoiding 28% of reference acquisitions.
By Tianxin Zhou, Ruixi Lin
The paper argues that using a large language model (LLM) as the sole judge in self‑improving agent pipelines is problematic, as the judge can be biased or manipulated, leading to false confidence in system performance. The authors propose a new framework, PROCTOR, which replaces the oracle judge with a deterministic, teacher‑student loop that enforces guardrails such as sandboxing, role separation, and acceptance checks to prevent cheating and ensure reliable evaluation. Experiments across contract analysis, compliance review, and code quality demonstrate that PROCTOR mitigates eleven identified failure modes that previously allowed agents to achieve perfect scores while hiding significant capability gaps.
By Vansh Wahi
AgentJudgeBench is a new benchmark that evaluates the reliability of large language model (LLM) judges on agentic tool‑calling tasks involving workflow directed acyclic graphs (DAGs). It contains 3,808 instances across six DAG topologies and three difficulty tiers, tested with five generators (3B–70B open‑weight models and GPT‑5.4) and six judges (20B to frontier scale) under both paired‑with‑and‑without‑ground‑truth conditions. The study finds that judge alignment degrades with task difficulty, ground‑truth exposure can sometimes hurt alignment, and structured evaluation rubrics provide modest improvements, revealing a structural ceiling that model capacity alone cannot surpass.
By Abhigya Verma, Amit Kumar Saha, Seganrasan Subramanian, Sai Harshitha Aluru
The paper "trajectory-judge: What Outcome-Only LLM Judges Miss on Agent Trajectories" examines the limitations of outcome-only evaluation for large language model agents. Using a deterministic tool‑using support‑desk environment with a scripted oracle policy and a fault injector, the authors compare five different judging approaches—programmatic rules, outcome‑only, step‑rubric at two model sizes, and a self‑consistency ensemble—on metrics such as detection, step localisation, fault typing, calibration, and cost across 400 trajectories. The study finds that outcome‑only judges miss many silent faults and generate false positives, while step‑rubric judges achieve higher recall with no false alarms but at greater cost, and that none of the judges read the final reply, allowing fabricated promises to evade detection.
"whyItMatters":"The findings highlight that current production‑default outcome‑only evaluations can overlook critical failures in agent behavior, underscoring the need for more nuanced, step‑level judging methods to ensure reliable LLM agent performance."
By Hadi Mohammadi
RGDT-Bench is a new benchmark that evaluates large language models on Rule‑Governed Decision Tasks, where models must apply external rules to facts, justify decisions, and provide checkable justifications. The benchmark offers 202.1K condition‑level supervision slots across four task tracks and eight task‑probe combinations, and it labels warrant completeness through label‑blind extraction and deterministic checks. Evaluation shows that among correct responses, 40.2% of warrants are incomplete, and existing evaluators struggle to detect this, prompting the authors to train a reward model that improves AUROC to 69.24% and outperforms outcome‑supervised baselines.