2026-09-03 科技日报

扫描 551 篇候选内容 · 覆盖 650 个信息源 · 纳入 534 篇

📌 今日要点

• Proactive cyber defense for governments and enterprises
• Introducing Gemini 3.8 Flash and 3.8 Flash Cyber
• Real-Time Intelligence with IBM Time Series Models on Confluent
• Quantifying User Behavior Patterns to Build Better Predictive Features
• This Python Library Can Run Pandas Workloads Up to 20x Faster
• 5 Real-World Applications of Agentic AI in Enterprise Automation

📝 科技简讯

• Proactive cyber defense for governments and enterprises — Google DeepMind
• Introducing Gemini 3.8 Flash and 3.8 Flash Cyber — Google DeepMind
• Real-Time Intelligence with IBM Time Series Models on Confluent — Hugging Face Blog
• Quantifying User Behavior Patterns to Build Better Predictive Features — KDnuggets
• This Python Library Can Run Pandas Workloads Up to 20x Faster — KDnuggets
• 5 Real-World Applications of Agentic AI in Enterprise Automation — KDnuggets
• Task-Specific Prompt with Global Context for Multi-Task Graph Pre-Training — ArXiv ML (cs.LG)
• REAL-Q: E2E LLM Quantization via Dynamic Gradient Descent — ArXiv ML (cs.LG)
• Convergence issues in Relational Concept Analysis based on AOC-posets — ArXiv ML (cs.LG)
• DISTAL: Distillation and Self-Supervised Pretraining for Structure-Agnostic Materials Property Prediction — ArXiv ML (cs.LG)
• ReNFT: Repairing Mode Collapse in Reward Post-Training via Internal Probability-Mass Recalibration — ArXiv ML (cs.LG)
• Attention Sensitivity Is Not Enough: Dissociating Attention-Level and Behavioural In-Context Learning under Fine-Tuning — ArXiv ML (cs.LG)
• RW-LoRA: Communication-Efficient Decentralized LoRA Fine-Tuning via Random Walks — ArXiv ML (cs.LG)
• Resolution has a new Agent Foundations team — LessWrong
• What is neuralese and why is it bad? — LessWrong
• Notes on “A global workspace in language models” — LessWrong
• BLOOM-WILT: on-policy examples of any LLM behaviour, from a one-line description and logits alone — LessWrong
• Releasing Benign Trajectories for MonitoringBench — LessWrong
• Proactive cyber defense for governments and enterprises — Google AI Blog
• ATV Big Air Tour turned 3 days of work into 3 hours with ChatGPT — OpenAI Blog
• Accessing OpenAI models on Amazon Bedrock from Australia with global cross-Region inference — AWS Machine Learning
• Modernizing and scaling support operations with generative AI on AWS — AWS Machine Learning
• How an AWS team detects dashboard content failures at scale using Amazon Bedrock — AWS Machine Learning
• From code to diagrams: Agentic architecture documentation with Amazon Bedrock AgentCore — AWS Machine Learning
• Trinity: Agentic AI-powered transition planning for students with disabilities — AWS Machine Learning
• ZimaBlue: Evolving Generalizable World Action Models through Scalable Video Pre-training — ArXiv CV (cs.CV)
• A Lagrangian View of Flow Matching — ArXiv CV (cs.CV)
• Distributed Implicit Harm: A Compositional Safety Blind Spot in MLLM-Based Video Moderation — ArXiv CV (cs.CV)
• 纯血国产！4B 模型越级打怪，杀入万亿赛道 — 新智元
• Gemini 3.8，明日登场！ — 新智元
• AI 替主人抢健身课，顺手删了陌生人的预约！太听话的 AI 才最可怕 — 新智元
• 刚刚，全球首台「万元级」高性能人形机器人诞生！比 Mac 还便宜 — 新智元
• OpenAI Astra 发布在即，全自动网络攻防能力已突破天际！ — 新智元
• SkyProduction 天工工作台：从剧本到成片，一套工作台把精品短剧创作真正跑起来 — 量子位
• 还在为大模型洗数据熬夜？蚂蚁拿下 VLDB 工业最佳论文，一套宽表搞定 35PB 语料，效率狂飙 5.6 倍 — 量子位
• 阿里更新旗舰模型 Qwen3.8-Max，前端编程能力跃居全球第一 — 量子位
• 前字节强化学习专家孙鹏博士加盟星尘智能，完善 Physical AI 全栈技术布局 — 量子位
• 企业级 Agent 落地样板间！百融硅基员工批量上岗，按结果领工资 — 量子位
• Claude 最强 Fable 5.1 发布！8 项屠榜，最高降价 45%，反蒸馏机制上线 — 量子位
• 香港首个真实开放场景服务机器人落地兰桂坊 — 量子位

📊 数据概览