诺贝尔物理学奖授予神经网络先驱 Hopfield 与 Hinton

ChatGPT 发布，对话式 AI 走向大众

Dr. Claw: An AI Scientist Workspace for Vibe Research

唐家三少痛批“AI 泔水”：网络文学成 AI 洗稿重灾区，48 小时能生成 500 万字

京东启动“京东物理 AI 加速计划”，打造全球最大物理世界运营中心

京东云宣布与摩尔线程规划建设 10 万卡 AI 工厂

中国联通与魅族合作发布“小魔方”AI 手机：4 英寸小方屏，主打全语音交互与端云协同

曹操出行与豆包联合推出 AI 打车服务，能一句话选车、设置空调温度等

AI 开始「互发微信」了，谁来管管它们？| AI 器物志

深度智控获宁德时代、沙特阿美战投等重磅加码，加速打造物理AI时代算力与能源底座

CosmoH2G: A Hand-to-Gripper Transfer Dataset and Baseline Method for Object Manipulation with Complex Spatial Movements

MOLE: Detecting Insider Threats in AI Agents

AI 降低了创造的门槛，B 站则想放大创造的回声

SceneMosaic: Efficient and Diverse Simulation-Ready Scene Generation via Hybrid Agentic Layout Evolution

MaxKernel: Agentic Kernel Generation for TPUs

ENEAS: Embedding-guided Neural Ensemble for Adaptive Segmentation

ShallowStream: Index Shallow then Answer Deep for Streaming Video Understanding

Real-World Knowledge-Guided Change Data Synthesis for Remote Sensing

Google I/O：Gemini 2.5 与全模态搜索重塑

Anthropic 开放 MCP 协议：AI 应用的 USB-C

AlphaFold 3 发布：AI 破解生命分子相互作用

AI 资讯站上线：内容开源 + 本地索引检索

OpenAI AI 训练版权纠纷案新细节：GPT-5 可代笔乔治 ·R·R· 马丁撰写《冰与火之歌》

抢单“物理外挂”被判赔 300 万元：最高法发布 9 件反不正当竞争典型案例，涵盖 AI、直播电商等新兴领域

奥之心新款 PEN 复古无反相机曝光：2040 万像素 M4/3 画幅，AI 识别对焦

Procedural Graphs: Self-Evolving Execution Structures for LLM Agents

NeoHorse-1: Towards Recursive Self-Improvement via Agentic Post-Training with Routing Harness

AuK Technical Report: An Open-Source Foundational Model for Speech Generation and Editing

Online Draft Co-Training for Speculative Decoding in Large-Scale, Long-Context RL Post-Training

Multi-Grid Post-Training for Long-Form Multi-Shot Video Generation

Ask Before You Optimize: Dynamic Pre-Formulation Clarification for Interactive Optimization

τ^τ-Bench: An Environment for End-To-End, Realistic Agent Construction

BeaconKV: Key-Value Cache Compression Guided by Beacon Queries for Efficient Large Reasoning Model Inference

DRACO: Fine-Grained Credit Assignment with Dynamic Rubrics for Long-Horizon Agent Training

One Editor, Many Edits: A Unified Training-Free Framework for Diverse Video Editing

AdaptVPR: Route-Aware Hard Positive Generation for Robust Visual Place Recognition

Safety for Whom? Boundary-Aware Self-Distillation for Controlled LLM Safety Refusal

What Else Needs Fixing? Exploring Cost-Effective Test-Time Compute for Revision Propagation in Artifacts Generated Through Conversation

RoboTok: An Internet-Scale Data Engine for Human Demonstration Retrieval and Dexterous Manipulation Learning

Let Confidence Change, Not the Prediction: Prediction-Preserving Repair for Post-hoc Calibration

A Common Measure of Communication for Speech Brain-Computer Interfaces

Verify Before You Distill: Prompt-Level Teacher Gating for On-Policy Distillation

OpenClaw 刷屏：个人 AI 助理的开源引爆点

Apple Intelligence 发布：端侧 AI 进入消费级

早报｜库克看到华为三星后，推动苹果加速研发折叠屏iPhone/小米中折叠「卖爆了」，首销增长310%​/《旅行青蛙・中国之旅》12月8日停运

蚂蚁百灵发布首个金融增强模型，AI开始进入真实投研工作流

办公 Agent 大乱战，新势力TeleAgent 凭什么坐上牌桌

孕期记忆力下降背后的生物学机制]

铠侠 CEO 否认与 SK 海力士深化合作：价格涨幅已经“够高了”，不会“为了市场份额而追逐市场份额”

Miles v0.1: Production-Level Post-Training