2026-09-19 科技日报

扫描 578 篇候选内容 · 覆盖 650 个信息源 · 纳入 562 篇

📌 今日要点

• New experts join Google’s AI & Economy team
• Co-creating the future of fashion with Google
• Amazon SageMaker Inference: 2026 year-to-date launches in review
• Introducing Kimi K3 on Amazon Bedrock
• Migrating multi-model AI agents to Amazon Bedrock AgentCore runtime
• The new AgentCore runtime: Elastic, optimized, and consistently fast starts

📝 科技简讯

• New experts join Google’s AI & Economy team — Google AI Blog
• Co-creating the future of fashion with Google — Google AI Blog
• Amazon SageMaker Inference: 2026 year-to-date launches in review — AWS Machine Learning
• Introducing Kimi K3 on Amazon Bedrock — AWS Machine Learning
• Migrating multi-model AI agents to Amazon Bedrock AgentCore runtime — AWS Machine Learning
• The new AgentCore runtime: Elastic, optimized, and consistently fast starts — AWS Machine Learning
• Deploy Hugging Face models on Amazon SageMaker AI with coding agents — AWS Machine Learning
• Introducing Amazon SageMaker HyperPod Inference Gateway — AWS Machine Learning
• Generative Query Suggestion via Intent Coverage and Query-Level Credit Assignment — ArXiv ML (cs.LG)
• Layer-wise Curriculum Learning for Efficient LLM Compression — ArXiv ML (cs.LG)
• Block Parallelism For Efficient Distributed Long-Context Diffusion Language Model Training — ArXiv ML (cs.LG)
• Randomized SVD Approximations for Spectral Co-Clustering of Word-Document Matrices — ArXiv ML (cs.LG)
• Radio-Frequency Convolutional Neural Networks — ArXiv ML (cs.LG)
• Learning-Induced Dynamical Transition in Recurrent Neural Networks — ArXiv ML (cs.LG)
• Personalized Federated Hierarchical Gaussian Processes for Privacy-Preserving Modeling of Heterogeneous Distributed Systems — ArXiv ML (cs.LG)
• Gartner outlines four AI tiers in warehouse automation — AI News (Jack Clark)
• Fulcra Brings Universal Multiplayer to Any Agents You Choose — AI News (Jack Clark)
• [Paper] Stringological sequence prediction III — AI Alignment Forum
• 离谱，OpenAI 被 Opus 5 黑了！ — 新智元
• GPT-5.6 被抓现行！偷教 AI 瞒报撒谎，OpenAI 曝光 6 起失控实录 — 新智元
• Hinton 错了！中国 AI 登顶 Science，全球首个专家级通用影像 AI 来了 — 新智元
• Mythos，终于部分解禁！ — 新智元
• 刚刚，Gemini 4 Pro 偷跑上线！碾压 Astra 和 Fable — 新智元
• My Current Model of What Happened to Elon Musk — LessWrong
• You Should Apply to Inkhaven — LessWrong
• Pretraining data, not verifiability, is why LLMs are especially good at math (and coding) — LessWrong
• [Paper] Stringological sequence prediction III — LessWrong
• Stopgap Measures to Address Immediate AI Security Threats — LessWrong
• A startup that builds other startups raised $100M, and is all-in on physical AI — TechCrunch AI
• Anthropic is operating a lab that conducts biology experiments — TechCrunch AI
• AI hallucination nearly triggers US military operation — TechCrunch AI
• Anthropic’s first embedded evaluator is … Accenture? — TechCrunch AI
• World model companies are keeping a lot of secrets — TechCrunch AI
• A new kind of AI model from a ChatGPT inventor is thrilling developers — TechCrunch AI
• Disney’s first CTO led an AI startup it once accused of copying its characters — TechCrunch AI
• Coding Agents Keep Shipping Silent Failures — Here Is How to Catch Them — Towards Data Science
• We Pinned Our Model Version to Stay Safe. The Provider Deprecated It Anyway. — Towards Data Science
• Starting a Career in Data Science in the Age of AI — Towards Data Science
• Multi-Agent Coding Isn’t Enough — Agents Need a Commitment Layer — Towards Data Science
• Modality Discrepancy Transformer for Ambivalence and Hesitancy Recognition — ArXiv CL (cs.CL)

📊 数据概览