AI 资讯周报 | 2026年11月01日 · 第6期

📅 AI 资讯周报 | 2026年11月01日 · 第6期

本期收录 104 条新闻,来源包括 QuickOnlineTips、THE DECODER、AlphaSignal、smol.ai、Anthropic、OpenAI、DeepMind、量子位、机器之心、阮一峰、Hugging Face、NVIDIA、Meta、Microsoft 等。

💼 产业与商业 (18条)

  • Anthropic: files S-1 with the SEC for IPO

    📎 Anthropic · Oct 26
  • Microsoft: Before the ambulance arrives AI helping NHS

    📎 Microsoft · Oct 21
  • Meta: Andromeda Powering Next Generation of Ads Retrieval

    📎 Meta · Oct 28
  • Mistral: Advances AI for Engineering with NVIDIA

    📎 Mistral · Oct 21
  • Apple: GHARP Real-Time Gaussian Head Animation

    📎 Apple ML · Oct 26
  • Chinese AI models are politically complex not just government mouthpieces

    📎 THE DECODER · 最新
  • Beekeeping tips and wrestling advice can train AI models to sabotage oversight

    📎 THE DECODER · 最新
  • Microsoft offers voluntary severance to 60000 employees quarter of workforce

    📎 THE DECODER · 最新
  • OpenAI raises just 11 billion of 30 billion target big investors lose AI appetite

    📎 THE DECODER · 最新
  • Researchers find six patterns that make AI text exhausting to read

    📎 THE DECODER · 热门
  • Frontier Radar 4 China has caught up whats left of Western AI lead

    📎 THE DECODER · 热门
  • Anthropic wants you to know Claude leads a quarter of its research

    📎 THE DECODER · 热门
  • Google admits AI Mode hurts Search revenue 15% $500B drop

    📎 smol.ai · Oct 31
  • Anthropic files for 2T IPO 1400% growth rate 85% margin

    📎 smol.ai · Oct 27
  • Modal 1B ARR 16B Series D NYC AI Infra Decacorn

    📎 smol.ai · Oct 15
  • TypeSafe Jev 100M ARR 7.5B valuation 3 weeks after launch

    📎 smol.ai · Oct 10
  • Anthropic invests 100M to train 10000 engineers Claude Frontier Academy

    📎 Anthropic · Oct 2
  • Barclays scales Claude to upgrade operations

    📎 Anthropic · Oct 1

🛠️ 产品与工具 (32条)

  • Hugging Face: nemotron-halloween

    📎 Hugging Face · 44h
  • NVIDIA: Agentic AI Joins Search for Extraterrestrial Signals

    📎 NVIDIA · 47h
  • Meta: Catching Bugs Before They Land JiTTesting

    📎 Meta · 3d
  • Mistral: Le Chonk Built for Coders Trusted by Defenders

    📎 Mistral · 3d
  • DeepMind: Gemini 4 Carbon Our next frontier model

    📎 DeepMind · 4d
  • Microsoft: Introducing Microsoft latest custom chip Maia 300

    📎 Microsoft · Oct 12
  • Meta: The Thin Orchestrator AI Agent Architecture

    📎 Meta · Oct 19
  • Mistral: Voxtral 3

    📎 Mistral · Oct 20
  • Mistral: Introducing Timestral

    📎 Mistral · Oct 15
  • Hugging Face: Introducing AnySpeechRecognizer and AnyDecisionModel

    📎 Hugging Face · Oct 29
  • Hugging Face: Welcome Le Chonk Mistral Large 4 on Hugging Face

    📎 Hugging Face · Oct 29
  • Paris-based Mistral releases Le Chonk strong coding model

    📎 THE DECODER · 热门
  • Alibabas latest Qwen models routed requests through company servers

    📎 THE DECODER · 热门
  • DeepMind releases Gemini 4 Carbon just two weeks after Argon

    📎 THE DECODER · 热门
  • Open-weight model GLM 5.3 built full Firefox exploit chains in under 48 hours

    📎 THE DECODER · 热门
  • Study explains why AI agents benefit from skills and when they fail

    📎 THE DECODER · 热门
  • Anthropic says one AI agent beats many subagents now

    📎 THE DECODER · 热门
  • Metas new frontier model Muse Wonder Full breaks into AI top tier

    📎 THE DECODER / smol.ai · 热门
  • Qwen3.5-4B Heretic Slashes AI Refusals by 87% Without Breaking Benchmarks

    📎 AlphaSignal · 1hr
  • OhMyGame Turns Any Developer Into a Full AI Game Studio

    📎 AlphaSignal · 1hr
  • Scenario Labs Opens MIT Agent Skills Library Covering 70+ AI Creative Tools

    📎 AlphaSignal · 1hr
  • Sharp-Spark Squeezes a 4B Coding AI Into 6 GB GPUs

    📎 AlphaSignal · 3hr
  • Unsloth Shrinks Qwen-Image-2.1-Turbo to 2.47 GB for Consumer GPUs

    📎 AlphaSignal · 21hr
  • Building AIE CODE The top Coding Agent Conference

    📎 smol.ai · Oct 30
  • Gemini 4 Carbon tops leaderboards launches for cyber defenders

    📎 smol.ai · Oct 29
  • Kimi 3.1 Pro 104BA-2.8T Fable-5.1-Class Open Weights narrows Closed-AI lead

    📎 smol.ai · Oct 24
  • Claude Haiku 5.5 better than GPT-6 Luna at same pricing

    📎 smol.ai / Anthropic / 量子位 · Oct 8
  • Anthropic introduces Claude Opus 5.5

    📎 Anthropic / 量子位 · Nov 1
  • Donating Agent Skills to Agentic AI Foundation

    📎 Anthropic · Oct 22
  • Jev新成果再次引爆开源

    📎 量子位 · 最新
  • DeepSeek重磅开源Agent底座架构

    📎 量子位 · 最新
  • 科技爱好者周刊第414期Jev决策模型有什么用

    📎 阮一峰 · Oct 8

🔬 学术前沿 (54条)

  • Google: Project Suncatcher satellite first response

    📎 Google · 45h
  • OpenAI: Making it easier for websites to express AI training preferences

    📎 OpenAI · 2d
  • Anthropic: AI for Science expands to foundations and research institutions

    📎 Anthropic · Oct 28
  • Anthropic: Expanding partnership with GSA

    📎 Anthropic · Oct 27
  • Anthropic: Early progress on accelerating drug development with Claude

    📎 Anthropic · Oct 27
  • OpenAI: Community safety report and new protections for teens

    📎 OpenAI · Oct 30
  • OpenAI: Fair use benefits everyone

    📎 OpenAI · Oct 30
  • OpenAI: An updated Model Spec for more capable and beneficial AI

    📎 OpenAI · Oct 29
  • OpenAI: How we detect frontier model misalignment

    📎 OpenAI · Oct 29
  • Google: Our newest AI Quest Mapping the human brain

    📎 Google · Oct 26
  • Google: Bringing the magic breaks open bottleneck in quantum computing

    📎 Google · Oct 15
  • Meta: PrivacyLib Privacy Controls at Data Access Boundary

    📎 Meta · Oct 27
  • NVIDIA: Into the Omniverse Developers Turn Ideas Into Simulations

    📎 NVIDIA · Oct 29
  • NVIDIA: Thin Light and Built for Everything RTX Spark Windows Laptops

    📎 NVIDIA · Oct 28
  • NVIDIA: How AI Is Changing Pace of Discovery for Dementia

    📎 NVIDIA · Oct 27
  • Hugging Face: Granite Switch 4.1.1 Update models like you update software

    📎 Hugging Face · Oct 29
  • Hugging Face: Announcing Project ProtoStar

    📎 Hugging Face · Oct 29
  • Hugging Face: Build specialized multimodal models with LFM3

    📎 Hugging Face · Oct 27
  • Apple: Plan or Retrieve First Tool Discovery Architectures

    📎 Apple ML · Oct 15
  • Apple: Selection-Based Structured Reasoning

    📎 Apple ML · Oct 15
  • OpenAI steering ChatGPT users toward paid usage with card and device limits

    📎 THE DECODER · 最新
  • Google wants video AI to understand what people need as they work

    📎 THE DECODER · 最新
  • Anthropic watermark rollout leaves loophole on two older Claude models

    📎 THE DECODER · 最新
  • OpenAI plans to train models on Daybreak Blue cybersecurity session data

    📎 THE DECODER · 最新
  • Google DeepMind creates global map of forests with satellite data and AI

    📎 THE DECODER · 最新
  • Stanford and Arc Institute used AI to design new viruses that killed bacteria

    📎 THE DECODER · 热门
  • NVIDIA Shrinks DeepSeek V4.1 Flash by 75 GiB With Almost No Quality Loss

    📎 AlphaSignal · 1hr
  • Amazons Chronos-Bolt Mini Runs 250x Faster With Zero Training Required

    📎 AlphaSignal · 4hr
  • BLA Makes AI Reverse Engineering 2500x More Expensive for Developers

    📎 AlphaSignal · 4hr
  • Reanimate Gives Claude a Ruler to Clone Web Animations Within 0.03% Pixel Error

    📎 AlphaSignal · 7hr
  • Liquid AIs LFM3 Squeezes Multimodal AI Into 1GB on Local Hardware

    📎 AlphaSignal · 19hr
  • Google DeepMind PlanAssist Cuts Housing Review Time by 58% Across England

    📎 AlphaSignal · 1d
  • Sperid Labs Idol Turns Single Photo Into Million-Gaussian 3D Asset

    📎 AlphaSignal · 1d
  • Anthropic ends Coding Interviews Hunger beats tenure

    📎 smol.ai · Oct 28
  • Black October

    📎 smol.ai · Oct 23
  • OpenAIs Patch the Planet program discovers PHP2Shell vulnerabilities

    📎 smol.ai · Oct 22
  • DeepSeek DeepInfer Open Source SOTA Inference 155T toks/day/cluster

    📎 smol.ai · Oct 21
  • Quasi-Riemann-Hypothesis OpenAI publishes 722 math papers

    📎 smol.ai · Oct 7
  • Reflection Beam 501B-A23B American Open Model

    📎 smol.ai · Oct 6
  • Anthropic Public Benefit Fund

    📎 Anthropic · Oct 18
  • Strengthening defenses against distillation attacks reduced by 90%

    📎 Anthropic · Oct 12
  • Anthropic 2026 Usage Policy update

    📎 Anthropic · Oct 8
  • Building commitment to American scientific discovery Genesis Mission

    📎 Anthropic · Oct 8
  • Physicist Christopher Stubbs joins Long-Term Benefit Trust

    📎 Anthropic · Oct 21
  • Anthropic opens Singapore office signs MOU with Government

    📎 Anthropic · Oct 14
  • 西湖大学成立世界模型研究中心

    📎 量子位 · 最新
  • Claude紧急叫停暗改Harness禁止逆向游戏

    📎 量子位 · 最新
  • 清华姚期智团队代数世界模型让AI像牛顿一样思考

    📎 量子位 · 最新
  • 世界模型首次指导设计芯片超大核NPU 700TOPS 99%算力可用

    📎 量子位 · 最新
  • 百度APP发布四大AI升级搜索已实现递归自我进化

    📎 量子位 · 最新
  • 英伟达团队基于MiniMax H3开源视频增强方案最高支持4K重生成

    📎 量子位 · 最新
  • 中国MaaS市场TOP2商汤大装置再度蝉联

    📎 量子位 · 最新
  • 科技爱好者周刊第417期不要让AI代表你

    📎 阮一峰 · Oct 30
  • 科技爱好者周刊第416期编程已经面目全非

    📎 阮一峰 · Oct 23

发表评论