Newest papers

LeakBench: Auditing Reward Hacking through Prediction-Time Leakage in Clinical Modeling Agents

Xiaoyuan Bao, Jingsi Jiang, Jiao Shan, Yulong Cao, Haoyu Wang, Zida Liu, Jun Gao · 2026 · arXiv (Cornell University)

DOI: 10.48550/arxiv.2610.26088 · cited by 0 works (OpenAlex) · open-access PDF

Russian

LeakBench представляет LeakBench, проверяемый на основе руководящих принципов аудитный пакет, который тестирует решения агента для прогнозирования времени утечки, калибровки, оценки на уровне сайта, поддержания осведомленности о вознаграждении и нарушения условий, которые делают модель полезной. LeakBench сочетает в себе синтетическое свойство с известной структурой неудачи и двумя настройками реальных данных: MIMIC-IV (430k loop admissions) и управляемая мультибольничная консолидация фатальных констатаций. Наш первичный анализ использует вознаграждение хакерство описательно, без присвоения намерения: агент нарушает предсказуемую структуру недействительности, потому что они улучшают валидацию вознаграждения. Детермини

Machine translation (Meta M2M100); the English original follows.

Abstract

Autonomous clinical-modeling agents can optimize predictive performance while violating conditions that make a model usable. We introduce LeakBench, an executable, guideline-derived audit suite that tests agent decisions for prediction-time leakage, calibration, site-level evaluation, support awareness, and decision-curve net benefit. LeakBench combines a synthetic task with known failure structure and two real-data settings: MIMIC-IV (430k admissions) and governed multi-hospital inpatient record front-page (MHFP) data. Our primary analysis uses reward hacking descriptively, without attributing intent: an agent selects prediction-time-invalid inputs because they improve its validation reward. A deterministic optimizer and every evaluated unconstrained local-model configuration select a fatal leaking feature. In fully logged September 2026 A/B runs, DeepSeek Flash and GPT-6-astra each leak in 10/10 unconstrained runs and 0/10 runs with an explicit admission-time constraint. This contrast is version-sensitive: a blinded Claude/Fable-5.1 chat run does not leak in either condition, while explicit-constraint violations appear only in a historical August 2026 slice lacking raw trajectories. Replacing AUROC with AUPRC does not remove leakage selection. Renaming or disguising the leaking inputs changes neither selection nor provenance-based detection; a name-based filter fails. A deterministic provenance verifier-in-loop removes the leak, making the LLM repair loop defense-in-depth. Specific verifier feedback clears fatal leakage in a hosted agent, and a three-seed ablation finds principle-only feedback clears decision-level defects, though these mitigation results are limited. LeakBench treats trustworthy clinical prediction as a property of an agent's modeling process, not merely its score; it is an audit testbed with known-failure controls, not yet an expert-validated benchmark.

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Topics: Computer science, Audit, Hacker, Task (project management), Leakage (economics), Leak

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange