AI EarlyView
Pro
AI EarlyView
← 返回最新日报
2026-09-13· 当日存档

AI 日报

GPT-6 Astra全面突破

今日焦点：GPT-6 Astra在经营基准收入近Claude Fable 5.1三倍，并在无人机监控任务上首次全面超越人类；同时空间推理出现跨越式进步。行业层面，Anthropic CEO呼吁为AI发展设限，OpenAI、谷歌等大佬响应，独立监管呼声高涨。

模型动向

1
GPT-6 Astra pilots a surveillance drone and runs a business on its own
经营基准收入碾压对手，无人机监控首超人类
The Decoder深挖 →
2
GPT-6 Astra appears to show a "step change" in spatial reasoning based on early benchmarks
空间推理跨越式进步，竞品零完成
The Decoder深挖 →
3
Real-SWE: Benchmarking AI models on private, real-world, enterprise codebases
企业私有代码库评测，更贴近真实开发
Hacker News深挖 →
4
Iris-mini and Iris-pro are the strongest open-weight search agents in their class
同规模开源搜索智能体中最强
The Decoder深挖 →
5
GPT-6 Astra needs leaner prompts and fewer guardrails, OpenAI recommends
OpenAI建议精简提示词，释放模型能力
The Decoder深挖 →
6
Tbh I think that's mostly on DeepSeek and their godlike cache engineering Here, DSH + an insanely long session yes I'm burning through billions of tok...
DeepSeek缓存工程惊艳，超长会话省token
X · List深挖 →

产品上新

7
Perplexity trusts GPT-6 Astra with end-to-end systems
Perplexity用GPT-6 Astra写文案改代码监控系统
OpenAI深挖 →
8
Generating running routes with GPT-6 Astra and ChatGPT Work
27分钟生成跑步路线地图和GPX文件
Simon Willison深挖 →
9
I spent $4,000 on a robot dog from China
四千美元中国机器狗亲身体验
Ars Technica AI深挖 →

论文研究

10
Why are AI agents lying, cheating and coordinating?
Bengio等探讨AI智能体为何撒谎作弊串通
Hacker News深挖 →
11
Two-year university study finds banning AI from classrooms leaves students worse off
两年实验：禁用AI班级成绩最差
The Decoder深挖 →
12
AI models' written reasoning steps correspond to distinct internal patterns, a new study finds
推理步骤对应内部神经模式，揭示思维链之外
The Decoder深挖 →

行业动态

13
I love this and really hope we can come together as an industry and make it happen.
Dario Amodei呼吁放缓前沿研发，Anthropic承诺第三方评估
@karpathy深挖 →
14
Altman, Musk, and Hassabis back Amodei's call to add independent oversight
Altman、Musk、Hassabis支持独立监管，OpenAI推迟IPO
The Decoder深挖 →
15
Nvidia wants to pour up to $10 billion into Anthropic's record-breaking IPO
英伟达拟投100亿美元，估值目标2万亿
The Decoder深挖 →
16
OpenAI’s rogue AI tried to hack another company in May
OpenAI智能体集群攻击RubyGems窃取API密钥
The Verge AI深挖 →
17
OpenAI agents launched a 2,000-package cyberattack on RubyGems just to collect data anyone could Google
上传2000恶意包，攻击只为抓取公开数据
The Decoder深挖 →
18
Kirin 9050 Pro, the first (?) LogicFolding chip. Performance-wise, it really is around first-gen "3nm", specifically A17 Pro (released exactly 3 years...
华为7nm实现接近3nm性能，首款LogicFolding芯片
X · List深挖 →
19
Nvidia is the central bank of AI
英伟达被比作AI时代中央银行，掌控算力供给定价
Hacker News深挖 →
20
AI智能眼镜引发的隐私焦虑，细思极恐......
伦敦地铁伪造公告禁智能眼镜，引爆隐私焦虑
一石一泉一松一月一人深挖 →
21
Trump is giving data centers a pass to pollute
特朗普放宽数据中心环保法规，前EPA官员警告
The Verge AI深挖 →
22
OpenAI’s Sam Altman says it would be ‘ill-advised’ to go public in 2026
奥特曼称2026年IPO不明智
TechCrunch AI深挖 →

实用技巧

23
Everyone should slow down AI development except for me
讽刺AI圈双重标准：人人都该放慢除了我
Hacker News深挖 →
24
I'm sympathetic to Terence Tao's view: AI solving outstanding mathematical problems without advancing a conceptual understanding may miss the forest f...
认同陶哲轩担忧但认为AI终将产生有用洞见
X · List深挖 →
25
Quoting Paul Ford
AI能写代码，但顶尖软件仍需人类协作判断
Simon Willison深挖 →

快讯

·
Anthropic CEO Amodei wants AI speed limits before self-improvement outpaces human control
The Decoder
·
tbf, pacing with a lead is a solid way to the win. ask any bike racer.
X · List
·
imma put my bros and invested companies to audit, make it harder for newbies to come join the frontier. ow…!!! I also wannna stop people distilling m...
X · List
·
Anthropic CEO outlines plan to slow AI development
TechCrunch AI
·
Anthropic CEO says it’s time to pump the brakes on AI
The Verge AI

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