AI检测器宣称1%误报率，但实际可能误判750名学生，这篇文章揭露大学面临的真实困境。
The detector said I cheated. I wrote every word myself. That's the opening line of a Reddit post from a public health student at the University at Buf…
一种新颖的攻击方法通过连续可控风格转换，成功绕过AIGC检测器，揭示其根本脆弱性
arXiv:2605.00924v2 Announce Type: replace-cross Abstract: AI-generated content (AIGC) detectors are increasingly deployed in high-stakes settings such…
揭示AI生成垃圾论文正严重威胁学术出版的可信度，科学界需警惕。
Article URL: https://www.theatlantic.com/science/2026/01/ai-slop-science-publishing/685704/ Comments URL: https://news.ycombinator.com/item?id=4843240…
「劝人别用AI」的文章竟是用AI写的，这波讽刺反转让媒体和学界集体破防。
IT之家 6 月 3 日消息，据英国《卫报》3 日（今天）报道，西悉尼大学负责质量与诚信的副校长凯丝 · 埃利斯教授 使用 AI 撰写评论文章 ，文章的主题是 劝学生认真完成学习任务，不要借助 AI 技术走捷径 。随后，刊登该文章的《悉尼先驱晨报》将其从网站撤下，并称这篇稿件“ 不可接受 ”。 埃利…
LLM驱动证据升级，自动验证科学主张与引文匹配，提升科研诚信。
arXiv:2605.27710v1 Announce Type: new Abstract: Misalignment between claims and their cited evidence is a common failure mode in reports generated by …
Nature发文警示：AI写科学综述存在根本性缺陷，可信度存疑
Nature, Published online: 26 May 2026; doi:10.1038/d41586-026-01616-3 The production of the highest-quality literature reviews requires the judgement …
超六成上班族“假装懂AI”只为晋升，调查揭示职场AI焦虑与“信心鸿沟”
IT之家 5 月 23 日消息，据外媒 TechRadar 今晚报道，GCheck 发布的最新数据显示，越来越多职场人正在“假装自己很懂 AI”：职场里正在出现所谓的“AI 信心鸿沟”：很多员工在公开场合表现得非常懂 AI，但私下其实 并不真正相信自己的 AI 能力 。 调查发现， 63% 员工承认…
揭露AI产品承诺与现实巨大差距：开源标签下藏着过时文档和悄然收紧的限制。
Article URL: https://iceworks.cc/blog/uctd/ Comments URL: https://news.ycombinator.com/item?id=48231555 Points: 1 # Comments: 0
ICML '24重量级案例：通过大规模监测揭示ChatGPT对AI会议同行评审的渗透与影响
arXiv:2403.07183v4 Announce Type: replace-cross Abstract: We present an approach for estimating the fraction of text in a large corpus which is likely…
Nature权威观点：AI正在威胁科学赖以生存的不确定性，引发科研诚信深层反思
Nature, Published online: 19 May 2026; doi:10.1038/d41586-026-01605-6 AI might jeopardize the uncertainty required in science
大规模实证分析揭示大语言模型时代引文造假现象，拷问学术信任根基
arXiv:2602.06718v2 Announce Type: replace-cross Abstract: Citations provide the basis for trusting scientific claims; when they are invalid or fabrica…
arXiv对AI生成伪科学内容零容忍，违规者将被封禁一年并永久标记，科研诚信再升级。
One of the site's moderators described the new policy on social media.
arXiv禁止作者一年内提交AI全权代写的论文，维护学术诚信，科研人需关注政策变化。
ArXiv is doing more to crack down on the careless use of large language models in scientific papers.
YC创始人的诚信危机引发热议，是少数个例还是创业文化变味？
Ask HN: Are all the YC founders lying offlate ? I've been seeing more and more examples of things like these where apparently YC founders are just dow…
AI正让大学成绩“通货膨胀”，谁受益？谁受害？
Article URL: https://www.axios.com/2026/05/16/ai-grade-inflation-college-classes Comments URL: https://news.ycombinator.com/item?id=48161402 Points: 5…
ArXiv重拳出击：提交AI伪造论文的作者将被封禁一年。学术打假从源头抓起，LLM生成内容必须由作者负全责——再也不让“再写200字摘要”这类机器痕迹蒙混过关。
Article URL: https://www.404media.co/new-arxiv-rules-ai-generated-papers-ban/ Comments URL: https://news.ycombinator.com/item?id=48159794 Points: 5 # …