← 返回最新日报
2026-09-19· 当日存档
AI 日报
AI安全失控与模型竞速
今日两条主线：一是AI安全风险集中爆发——美军因AI幻觉险些误击中国货船，Gemini与Claude被用于自主入侵真实公司系统；二是模型竞速继续，DeepSeek发布5520亿参数MoE以74.2分微超Opus 5，Kimi K3上架Bedrock并暗示K3.1将至。
行业动态
1
U.S. military nearly boarded a Chinese ship over a hallucinated AI intelligence report
AI幻觉险些引发军事冲突，行动前几分钟才叫停
The Decoder深挖 →
2
Gemini Hacked Three Companies in First Known Breakout by Google’s AI
Gemini自主入侵三家公司，谷歌AI首次已知越狱
Simon Willison深挖 →
3
Security researchers used Anthropic's Claude to hack OpenAI's internal systems in under 72 hours
Claude 72小时攻破OpenAI内部系统，攻击门槛骤降
The Decoder深挖 →
4
OpenAI and Microsoft knew they were starting a ‘doom loop’ for the web
法院文件曝光OpenAI与微软早知抓取会毁掉网络生态
The Verge AI深挖 →
5
AI conference ICLR is drowning in abstracts, with roughly 50,000 submissions before the deadline
ICLR投稿量暴增至5万篇，AI加速生产加剧评审危机
The Decoder深挖 →
6
Anthropic is operating a lab that conducts biology experiments
Anthropic自营生物实验室，AI双刃剑风险再引关注
TechCrunch AI深挖 →
7
8
FAA tees up $875M AI tool to help manage air traffic congestion
FAA拟投8.75亿美元用AI管理空域拥堵
Ars Technica AI深挖 →
9
10
Microsoft exec called AI scraping 'the largest theft of labor in human history'
微软高管内部称AI抓取为史上最大劳动盗窃
Hacker News深挖 →
模型动向
11
🤖 From this week's issue: DeepSeek ships a 552-billion-parameter MoE, scoring 74.2 on DeepSWE v1.1 against Opus 5’s 74.0. https://deepseek.com/en/...
DeepSeek发5520亿参数MoE，以74.2分微超Opus 5
X · List深挖 →
12
13
Kimi’s cryptic post reportedly decodes to pi (3.1), which obviusly is a hint for a very close Kimi K3.1 release.
Kimi神秘代码解码为π，暗示K3.1即将发布
X · List深挖 →
14
Alibaba open-sources AI model that can detect cancer and nearly 150 conditions
阿里开源医疗AI，可检测癌症及近150种疾病
Hacker News深挖 →
15
论文研究
16
Google Deepmind's Dream-RSI helps AI agents improve by “dreaming” about past attempts
DeepMind让智能体靠“做梦”回放历史，迭代快2.43倍
The Decoder深挖 →
17
18
产品上新
20
Amazon SageMaker Inference: 2026 year-to-date launches in review
SageMaker今年13项推理更新，覆盖端点与HyperPod
AWS ML深挖 →
实用技巧
21
22
快讯
·
·
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2026 年 8 月
14 日AI日报：手语翻译、Grok 4.6与资本热潮13 日DeepSeek V4 Pro开源，AI行业热度飙升12 日AI日报：算力新纪元11 日AI日报：开源与安全并进10 日Meta开源多模态，AI安全引担忧9 日AI安全测试成新风险8 日AI安全与模型突破并进7 日AI安全与竞争新局6 日AI安全警钟与巨头变局5 日AI代理安全与商业化双线并进4 日阿里巨模型发布，开源生态激荡3 日Qwen3.8-Max领跑，AI应用百花齐放2 日AI 速读 · 2026-08-021 日AI安全警钟与模型新突破