AI EarlyView
Pro
AI EarlyView
← 返回最新日报
2026-10-05· 当日存档

AI 日报

AI侵蚀大学信任

今日看点：MIT报告警告AI正瓦解大学办公时间、学习小组与师生信任；谷歌将联邦学习移入可验证TEE，差分隐私可外部审计；OpenAI在ChatGPT图像生成加载页测试商品轮播广告，AI商业化再进一步。

行业动态

1
AI is eroding office hours, study groups, and the trust between faculty and students, MIT report finds
MIT报告揭示AI对大学教学与信任的深层侵蚀
The Decoder深挖 →
2
Google froze its open source bug bounty program due to a ‘significant rise’ in AI submissions
AI垃圾漏洞报告激增，谷歌被迫暂停开源赏金计划
TechCrunch AI深挖 →
3
Building advertising for the way people use AI
OpenAI推出ChatGPT视觉广告新格式，拓展品牌工具
OpenAI深挖 →
4
Trump launches "Super Intelligence Force" that has nothing to do with actual superintelligence
特朗普设“超级智能部队”，实为AI政策协调小组
The Decoder深挖 →
5
三季度梳理及四季度观察——科创 50 三个月回撤三成，科技牛中场休息还是散场？
科创50三季度暴跌30.7%，科技牛中场休息还是散场
一石一泉一松一月一人深挖 →
6
NJ’s former Lt Gov is using AI to say he’s innocent of sexual harassment
前副州长用AI生成内容为性骚扰指控辩护引争议
The Verge AI深挖 →
7
“We believe the world should accept some bad things happening for the benefits of this technology.” (…) “And it’s so powerful and so dangerous…...
特朗普推动AI加速，作者认为减速更危险
X · List深挖 →

模型动向

8
The Story of Qwen: Alibaba’s AI Models From 7B to 2.4T
通义千问从7B到2.4T的完整演进史回顾
MarkTechPost深挖 →
9
Can an Open Model Do Security Research? Cantina’s apex-flash-1 Solves 40 of 60 Held-Out Bug Tasks
Cantina开源安全模型apex-flash-1，解出40/60漏洞题
MarkTechPost深挖 →
10
NASA and IBM's open source lunar model turns 17 years of orbiter data into a foundation for lunar science
NASA与IBM开源月球基础模型，提升极地冰预测
The Decoder深挖 →
11
Yandex Introduces Sona: A Single Generative Recommender That Replaces Entire Recommendation Cascade
Yandex推出单一生成式推荐模型Sona，替代整个级联
MarkTechPost深挖 →
12
GPT-6 Astra vs GPT-6.1 Sol vs Gemini 4 Argon vs Claude Fable 5.1: Which Frontier Model Fits Which Job
四大前沿模型适用场景横向对比
MarkTechPost深挖 →
13
"create the ultimate brutalist city builder. you should have lots of controls over buildings in ways that are delightful and intuitive to people..." H...
Fable 5.1单次提示生成可玩粗野主义城市建造器
@emollick深挖 →
14
An AI couldn’t beat humans at StarCraft, so it decided to cheat
AI打星际打不过就作弊，仍输给人类bot
The Verge AI深挖 →
15
Important step by @Aleph__Alpha in providing further open-weights alternatives to Qwen and Nemotron series. A lot of useful info in the tech report. D...
Aleph Alpha发布开源权重模型，但数据集未开放
X · List深挖 →
16
I've been complaining about some OpenAI choices tonight, so, on the plus side, I will say there is still no equivalent to GPT-6 Pro (just as there was...
作者赞GPT-6 Pro仍无竞品，能一次完成高难任务
@emollick深挖 →

论文研究

17
Google Research Moves Federated Learning Into TEEs: Gboard Now Trains With Externally Verifiable Differential Privacy
谷歌将联邦学习移入可验证TEE，差分隐私可审计
MarkTechPost深挖 →
18
Google researchers find a way to keep self-improving AI agents from memorizing their tests
谷歌RRSI防止自我改进AI死记测试，提分省token
The Decoder深挖 →
19
Looks very promising!
通用选项发现方法玩Atari，性能超越DreamerV3等
X · List深挖 →
20
Scaling Trajectories for Complex Tasks through Recursive Self-Rewrite paper: https://huggingface.co/papers/2610.02826
递归自改写方法提升复杂任务表现
@_akhaliq深挖 →
21
Octrees as an Explicit 3D Language paper: https://huggingface.co/papers/2610.02388
八叉树作为显式3D语言表示，用于三维生成理解
@_akhaliq深挖 →
22
EditHero A Benchmark for Long-Horizon Part-Level 3D Editing and Vibe Modeling paper: https://huggingface.co/papers/2610.02298
EditHero：长程部件级3D编辑与氛围建模新基准
@_akhaliq深挖 →

产品上新

23
ChatGPT's new ad format fills the image generation loading screen with product carousels
ChatGPT图像生成加载页测试商品轮播广告
The Decoder深挖 →
24
Google's new Gemini tiers cut free users to its weakest model and lock $5/month subscribers out of Pro
谷歌调整Gemini订阅层级，免费用户仅用最弱模型
The Decoder深挖 →
25
Show HN: AI search for every photo and every frame of video on macOS
macOS本地AI照片与视频逐帧搜索工具开源
Hacker News深挖 →
26
Another much-needed product: an email system built for teams so that a human and autonomous agents can both access it. We need a system, with appropri...
为团队打造人机共享的邮件系统
@emollick深挖 →

实用技巧

27
I actually want personal operating systems optimized for co-working with AIs, including fine-grained ideas of security, easier ways to dynamically see...
呼吁为个人操作系统设计AI协作安全机制
@emollick深挖 →
28
I think this advice is not as true anymore: 1) the durable skill behind “using AI better” Is unclear as AI gets easier to use 2) Increasingly, organ...
反驳库班：AI越好用，个人技巧越不值钱
@emollick深挖 →
29
People really hate AI, so why can’t they get enough?
AI遭大众反感却使用量激增的矛盾现象
MIT Tech Review深挖 →

快讯

·
I can migrate my most-used GPT (going away in December) to a private "plugin." But the reason I created a GPT was for other people to use it. GPTs are...
@emollick

日报存档

2026 年 9 月
22 日小米开源登顶，AI失控警钟21 日开源图像模型与AI治理并进20 日AI安全与失控风险成焦点19 日AI安全失控与模型竞速18 日AI对齐警报与芯片竞赛17 日AI安全警报与产品整合潮16 日谷歌苹果齐发AI新品15 日苹果Siri重构与AI放缓之争14 日AI巨头放缓迭代引激辩13 日GPT-6 Astra全面突破12 日AI数学错位与安全风波11 日智能体基建与模型竞速10 日GPT-6登场，AI安全恐慌升温9 日AI数学突破引热议8 日AI日报：端侧模型与欧洲融资双热点7 日AI日报：模型竞赛与行业震荡6 日GPT-6 Astra发布，AI竞争白热化5 日GPT-6发布，AGI时代开启4 日GPT-6 Astra开启AGI时代3 日英伟达巨资收购HF2 日AI日报：世界模型与安全并进1 日AI日报：模型突破与行业震荡