arXiv 2609.29429 · cs.AI · 2026-09-24
Just Ask Jev: Reinforcement Learning for Calibrated Decisions as a Zero-Shot Detector of AI Alignment Failures
Ruoqi Guo, Yi Liu, Gelei Deng, Yuekang Li, Lida Zhao, Yutao Wu, Simin Chen, Ying Zhang 等 9 人
机构待核实
在最新榜 #170 本站收录于 2026-10-21
一句话要点
测试经 RL 校准的决策模型 Jev 能否零样本检测谄媚、越狱、欺骗等十类对齐失败。
Tests whether Jev, a calibrated decision model, can zero-shot detect ten alignment failures such as sycophancy, jailbreaks and deception.
阅读 arXiv 原文 PDF Hugging Face 讨论 代码仓库 项目页
主题与学科
同主题：安全与可解释
- Behavioral Compositionality of Sparse Autoencoder Representations ▲110
- DecepEval: A Benchmark for Evaluating Deception in LLM Agents ▲85
- Lexical Reference Frames: Interpreting LM Residual Streams in Native Semantics ▲55
- U-Space: Uncovering When and Why Uncertainty Arises in Language Models ▲54
- APEX-Harm: Evaluating Agent Safety in Shared Professional Workspaces ▲26
数据说明
元数据来自 arXiv API（已核实）；点赞来自 Hugging Face 每日论文；引用来自 Semantic Scholar。最近刷新 2026-10-22。
要点由本站根据摘要自写，不转载摘要原文，请阅读原文。