揭露LLM编码工具压缩历史时，可能凭空捏造已终止进程的确认结果，引发对AI工具可靠性的深刻反思。
arXiv:2607.13071v1 Announce Type: cross Abstract: Agentic LLM coding tools compress long session histories into compaction summaries that subsequent s…
一篇实证研究，揭示大语言模型在生成代码时对自己安全性的认知偏差，关乎代码安全可靠性。
arXiv:2606.31159v1 Announce Type: cross Abstract: Large Language Models (LLMs) are rapidly transforming software development, yet their use in securit…
新论文揭示LLM排序漏洞：两阶段token优化可操控排名，可靠性存疑
arXiv:2510.06732v2 Announce Type: replace Abstract: Large language models (LLMs) are increasingly used as rerankers in information retrieval, yet thei…
LLM作为对话裁判时会被表述框架误导，新框架DialDefer可检测并缓解这种偏差。
arXiv:2601.10896v2 Announce Type: replace Abstract: LLMs are increasingly used as third-party judges, yet their reliability when evaluating speakers i…
LLM可靠性新突破：聚类自评估方法简单有效，无需额外训练即可量化不确定性，ACL 2026研究成果。
arXiv:2606.03846v1 Announce Type: cross Abstract: Large language models (LLMs) demonstrate remarkable performance across diverse tasks, but they often…
AI基准评估面临污染威胁，论文揭示现有检测工具在分布偏移和大规模场景下严重失效，挑战传统审计可靠性。
arXiv:2606.03305v1 Announce Type: new Abstract: Benchmark contamination, where evaluation examples appear in a model's training data, threatens the va…
揭秘大语言模型「不确定性」的来源，一项严谨的技术评估论文，帮你理解LLM为何「不知道」。
arXiv:2604.10495v2 Announce Type: replace Abstract: As Large Language Models (LLMs) are increasingly deployed in real-world applications, reliable unc…
从理论不可能到局部补丁，这篇论文系统解析了LLM错误架构，为提升模型可靠性提供全新视角。
arXiv:2605.30628v1 Announce Type: cross Abstract: Universal LLM reliability is not a finite-library problem: across all possible tasks, tools, schemas…
Kalai和Vempala提出的概率框架，为大语言模型幻觉现象提供了严谨的形式化定义与校准条件，是理论突破。
arXiv:2605.26808v1 Announce Type: cross Abstract: Hallucination is a central limitation of large language models (LLMs), and substantial effort has be…
多轮对话让LLM可靠性下降？研究揭示在医疗等高风险场景中静态评测与真实使用的巨大鸿沟。
arXiv:2603.11394v3 Announce Type: replace-cross Abstract: Large language models (LLMs) excel on static benchmarks, but their performance across multi-…
AI代理依赖管理是门学问，封闭环境能避免库更新导致LLM行为突变，保护你的代理“认知不崩塌”。
Most developers treat AI agents like traditional software utilities. They run a quick pip install , throw some environment variables into a .env file,…
基于LLM隐藏状态探测的多语言幻觉检测新方法，有效应对非英语场景的可靠性挑战。
arXiv:2605.24919v1 Announce Type: new Abstract: Hallucinations in Large Language Models (LLMs) represent a critical barrier to their reliable deployme…
基于交互式证明理论，让大模型通过“证明者-验证者”相互校验，实现有选择地预测——知道何时该回答、何时该存疑。
arXiv:2605.25133v1 Announce Type: new Abstract: Reliably knowing when a language model is correct is almost as important as being correct. We introduc…
提出双重校准方法，同时校准LLM的知识与推理置信度，为提升模型可靠性提供全新路径。
arXiv:2601.11956v2 Announce Type: replace Abstract: Reliable reasoning in Large Language Models (LLMs) is challenged by their propensity for hallucina…