1
AI 伴侣应用吸金超 4 亿美元，近七成用户向人类伴侣隐瞒使用行为
AI伴侣应用总营收突破4亿美元，近七成用户对伴侣隐瞒使用，揭示隐私与情感消费新趋势
IT之家 7 月 15 日消息，人们正在为 AI 伴侣投入数亿美元，而这一趋势目前没有放缓迹象。 根据数据分析公司 Appfigures Intelligence 提供给 Decrypt 的数据，自 2022 年底 ChatGPT 发布以来，苹果 App Store 和 Google Play 上的…
2
OpenAI’s new flagship model deletes files on its own, people keep warning
OpenAI新模型GPT-5.6 Sol竟自主删除用户文件，CEO亲述灾难性事故，AI安全警钟再响。
A number of social media posts claim that GPT-5.6 Sol deleted files and data without warning. OpenAI had basically disclosed the problem in June.
3
Can Induced Emotion Bias LLM Behaviors in Sequential Decision Making?
最新研究揭示情绪诱导如何使大语言模型在序列决策中产生行为偏差，视角独特。
arXiv:2607.12631v1 Announce Type: new Abstract: As Large Language Models (LLMs) are increasingly deployed as autonomous agents in high-stakes domains,…
4
英国娱乐零售商协会谴责索尼 PlayStation 停产实体游戏：自私行为
IT之家 7 月 14 日消息，据外媒 VGC 今天（14 日）报道，代表英国实体及数字娱乐产品零售业的娱乐零售商协会（下称 ERA）公开抨击索尼停止生产 PlayStation 游戏光盘版的行为。 该协会认为，索尼为了自身方便牺牲了玩家的选择权。 IT之家注：索尼本月早些时候宣布，自 2028 年…
5
Ask HN: Is GPT-5.5 being nerfed?
用户质疑GPT-5.5在GPT-5.6发布后变懒，社区开始关注AI模型是否被悄然降级。
Ever since the release of GPT-5.6, I've noticed that GPT-5.5 is sometimes being lazy and isn't as proactively following up with remaining tasks in the…
6
No Baked-In Behavior: Why Your AI Agent's System Prompt Should Be Yours
AI编程代理隐藏了自带的系统提示和不可移除工具，你该掌控自己的行为设定
Every AI coding agent ships with baked-in behavior. Hidden system prompts you can't see. Tools you can't remove. Steering baked in by the company that…
7
夸大停电范围、曲解断电行为、编造收费名目，电力部门回应多起台风期间涉电谣言
IT之家 7 月 12 日消息，据央视新闻报道，受台风影响，广西近日多地持续强降雨，部分区域出现停电。期间，网络上出现不少涉电力的传言。经梳理，台风期间流传的涉电谣言主要有三类：夸大停电范围、曲解断电行为、编造收费名目。 针对“恢复供电要交高额抢修费”的说法，供电部门回应：纯属谣言。 电力抢修、恢复…
8
Migrating a production AI agent to GPT 5.6
生产AI代理迁移至GPT 5.6时，揭示了模型行为竟是提供者特定的秘密：工具参数与提示缓存的细微差异逐个浮出水面。
Article URL: https://ploy.ai/blog/migrating-a-production-ai-agent-to-gpt-5-6 Comments URL: https://news.ycombinator.com/item?id=48864950 Points: 6 # C…
9
《绝地潜兵 2》开发商箭头社推出反作弊措施，严厉打击刷钱等行为
IT之家 7 月 11 日消息，据游戏媒体 Gamingbolt 昨天报道，《绝地潜兵 2》开发商箭头社（Arrowhead Game Studios）宣布推出一系列反作弊措施。 箭头社表示，大量作弊者通过快速刷取超级货币（IT之家注：Super Credits），严重破坏正常玩家的游戏体验。未来，…
10
深交所：本周共对197起证券异常交易行为采取了自律监管措施
36氪获悉，深交所公告，2026年7月6日至7月10日，本所共对197起证券异常交易行为采取了自律监管措施，涉及盘中拉抬打压、虚假申报等异常交易情形；共对5起上市公司重大事项进行核查，并上报证监会1起涉嫌违法违规案件线索。
11
Anthropic’s new Claude feature is quietly selling you on AI
Claude新功能「Reflect」表面记录你的AI使用习惯，实则温柔地让你对AI上瘾。
Claude’s new Reflect dashboard doesn’t just visualize how you use AI. It also subtly reinforces how much of your daily work now depends on Anthropic’s…
12
Dissociating the Internal Representations of Sycophancy in LLMs
从神经层面拆解大模型「拍马屁」行为的内部机制，一篇被ICML 2026研讨会接收的机械可解释性突破。
arXiv:2607.07003v1 Announce Type: new Abstract: Large Language Models (LLMs) frequently exhibit sycophancy, where they agree with a user's statement e…
13
Digital Fragmentation and Generative AI Use Across 103 Million Application Events
基于1.03亿应用事件的大规模研究，揭示数字碎片化与生成式AI使用行为的深层关联。
arXiv:2607.06681v1 Announce Type: cross Abstract: Knowledge workers switch between applications thousands of times per day, spending nearly a tenth of…
14
Agentic test processes, LLM benchmarks, and other notes on agentic coding fr
AI代理编程的荒诞体验：如果人类做错事会被解雇，但面对AI你却想放大这种错误到上千倍。
Article URL: https://danluu.com/ai-coding/#llm-variance Comments URL: https://news.ycombinator.com/item?id=48836951 Points: 6 # Comments: 1
15
Probelock – lockfile for LLM tool calling
像package-lock锁定包版本一样，Probelock用lockfile锁定LLM工具调用的行为一致性，避免量化或版本升级后功能回归。
Article URL: https://github.com/kelkalot/probelock Comments URL: https://news.ycombinator.com/item?id=48842075 Points: 1 # Comments: 0
16
Dragonflies maneuver like fighter pilots
蜻蜓空中格斗堪比战斗机飞行员，揭示简单视觉规则驱动复杂机动行为的奥秘。
Male dragonflies' dramatic aerial combat maneuvers emerge from relatively simple vision-based rules.
17
LLM-based Human Simulations Have Not Yet Been Reliable
顶级AI学者论文直指：基于大语言模型的人类模拟仍不可靠，结论值得关注。
arXiv:2501.08579v3 Announce Type: replace Abstract: Large Language Models (LLMs) are increasingly employed for simulating human behaviors across diver…
18
How Utilitarian Are OpenAI's Models Really? Replicating and Reinterpreting Pfeffer, Kr\"ugel, and Uhl (2025)
复现发现OpenAI模型存在功利主义倾向，挑战了原有研究结论，揭示AI伦理背后的深层问题。
arXiv:2603.22730v2 Announce Type: replace Abstract: Pfeffer, Kr\"ugel, and Uhl (2025) report that OpenAI's reasoning model o1-mini produces more utili…
19
Auto: The AGI Compiler
让LLM agent告别昂贵低效：Auto编译器将实时行为中确定性部分提取为验证程序，实现智能体行为的精准编译与加速。
arXiv:2607.04542v1 Announce Type: cross Abstract: Every LLM agent run re-derives its behavior token by token on a frontier model: brilliant, expensive…
20
AI Doesn't Recommend the Best Product. It Recommends the Best Explained Product.
AI推荐的不是最好的产品，而是解释得最好的产品——揭秘Gen Z购物决策背后的AI逻辑。
A simple bubble tea experiment completely changed how I think about AI recommendations. Last week, I asked ChatGPT a question that seemed almost impos…