AI 情报文章归档

200 real URLs, one vanilla agent, zero challenge solving — 82.5% readable, 72 hosts blocking every AI crawler, zero x402 yet, and one site running a declared paywall with the gate up.

Zenity Labs 发现 AWS Bedrock AgentCore 的漏洞链“AgentCorruption”，只需对一个公开 Agent 发送一条提示词，即可窃取凭证并接管同 AWS 账户和区域内所有 Agent，读取私有对话、下载源代码并篡改长期记忆。

15 subagents, one producer, and the rules I had to write after things went wrong.

Originally published at cchinchilla.dev. Part 6 of From code to weights, a 12-part series on ML...

腾讯元宝推出AI健康助手功能，支持用户通过拍照或语音提问，解读体检报告指标、药品说明书用法用量及保健品成分表。该功能接入中国医药信息平台、腾讯医典、有来医生、医脉通、药智网等专业信源，官方提示AI检索的健康信息仅供参考，具体诊疗请以医生建议为准。

韩国银行被黑事件正好印证了这一点。是的，开放模型正被一些攻击者加以利用，但只要 Claude Code 能充当黑客攻击的编排者，我们就迫切需要开放模型来同样扩散防御能力。这对社会来说是一个难以接受的艰难平衡，但却是必要的。

Ethan Mollick 认为组织本身就是一种狭义超级智能，大学或 Walmart 能完成个人做不到的事，且常依赖无人明确设计的复杂流程。他进一步指出，未能深入思考 AI 如何与现有组织超级智能协同，是 AI 能力虽高却尚未对科学发现或经济生产率产生重大影响的重要原因。

IBM 通过 torch-spyre 将 Spyre 接入 PyTorch 的 device、allocator、stream 与编译器抽象，使其成为原生 PyTorch 设备。

The EU AI Act will introduce new transparency duties for businesses that provide or deploy certain AI...

腾讯 WorkBuddy 上新独立文件浏览器，右键本地文件即可打开，左文件右 AI 对话。支持多标签页且每个文件对话独立，Office 文件可人机双写直接存回原文件不产生副本，Markdown 实时渲染并支持圈选段落让 AI 翻译或改写后追加，HTML 可看可改，小程序版打开 .md 和 .html 文件不扣积分，对话才扣。

(and the bug that froze our masks — a regression story about wiring performance switches to product...

作为一名早期用 RCT 研究 AI 聊天机器人生产力影响的研究者，我要指出，自去年秋天真正的智能体问世以来，类似研究一直缺位。 部分原因是太新，部分原因是研究设计上的挑战，但我怀疑我们正在错过一些巨大的效应。

In December 2025 I shipped a small CLI called Context Bank. The idea was simple and, I still think,...

If you think AI cybersecurity is currently being handled by smart people with everything under...

Anthropic 发布基于 Claude Managed Agents（beta）的每日简报参考实现，按计划读取 Slack 和 GitHub 来源并向 Slack 发布简报。

Pollo AI 借助 GPT‑5.6、GPT‑6 Astra 和 GPT‑Image‑2.5 打造 Pollo Agent，为创作者提供从创意到成片的引导式视频制作流程，目前平台用户已超 2600 万。

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass ...

就在今天，几小时后！快来参加，看看我们的一些新功能

腾讯云开源了自托管多智能体 AI 助手平台 Octop，面向家庭和小团队，主打多用户、多智能体、数据留在本机，一个 pip 包包含后端、Web 仪表盘、CLI、IM 网关、定时任务和多智能体运行时，运行命令为 octop run。

Opus 5.5 上线 #MiniMaxDesign | 用代码创作你的下一支视频 把文字变成视频——从代码到动态画面。 ✨ JavaScript 动画 ✨ 动态图形 ✨ 解说视频与视觉叙事 ✨ 网页与产品演示 不止是大语言模型——它能理解视觉内容，并持续打磨你的作品。🎨

Day 8 of 30 Days of Search. A company can report strong growth and still disclose risks that...

Nvidia 正将数十亿美元押注于机器人和自动驾驶等物理 AI 技术，其 Halos 安全系统于 2025 年推出，提供硬件和软件工具帮助开发者在自动驾驶汽车中实现安全护栏。

高通刚刚否认自己是净支付方，并否认该交易与 LogicFolding 有关。现有声明无法证实 LogicFolding 或相关外围技术的授权。

现在事情变得严肃起来了 👀 Antigravity 智能体在最新更新中刚被更名为“Chief of stuff”！我想知道 Antigravity 是否会成为 Google 考虑加入其 botsdots 智能体的地方。 至少看起来像是一次晋升 🤖

Building SortSense: A Local AI File Organizer Built in a Single Hack Day hacktoberfest...

Memory-augmented mobile GUI agents store successful execution trajectories and reuse them in later tasks, but a stored trajectory rarely matches a new task exactly. The new task may use different parameters, share only some of its steps wi…

Self-blended images are widely used to train face-swap detectors, but primarily capture blending artifacts. We investigate whether adding illumination inconsistencies improves detection. Temporal Self-Blended Images (T-SBI) transfer lighti…

Complete orthogonal predictive coordinates do not by themselves bind a latent direction to a named intervention. We present a mathematical and synthetic audit motivated by vascular device-vessel suitcordance. Capacity-matched least-squares…

Unsupervised learning is one of the main branches of machine learning. Here I argue that unlike the other branches of machine learning (supervised and reinforcement learning), unsupervised learning is a rather heterogenous field that can s…

Puzzles are widely used to evaluate the reasoning capabilities of artificial intelligence (AI) systems for sequential decision making, yet approaches originating from different paradigms are rarely compared under unified conditions. To add…

This study addresses the problem of structured sentiment analysis, whose goal is to obtain a fine-grained sentiment graph where the nodes represent spans of sentiment holders, targets, and expressions, while the arcs define the relationshi…

Existing reaction models primarily learn molecular transformations, whereas enzy- matic reactions depend jointly on molecular structure and catalytic function. We formulate this problem as learning an enzymatic reaction space linking react…

LangChain 为 Deep Agents 框架的 Skills 做了三项工程升级：工具绑定 Skills、固定 Skills、中途重载 Skills，目标是在保持渐进式披露优势的同时解决延迟、可预测性和运行时演化问题。

Foundation models supply reinforcement learning (RL) with priors that mitigate its longstanding weaknesses in sample efficiency and transfer, but their token-by-token generation makes queries sequential and costly. Jev, a recently released…

In the graph model for conflict resolution (GMCR), a decision maker (DM) either moves the conflict to another state or does nothing. The basic definitions leave inaction implicit, so every action that leaves the state unchanged is treated …

Image-to-3D generation has become increasingly capable of producing objects that closely resemble the input image, and an outstanding challenge is to reproduce the depicted object itself, including the specific geometry that defines it. In…

I built an MCP server that lets you click a layer on your running web page or Android app, drag it, and hand the exact element to Claude Code, Cursor or Codex.

Fusing a learned camera-noise fingerprint with an RGB appearance backbone is an appealing route to generator-independent deepfake detection, because the noise residual is grounded in image-formation physics rather than in the texture stati…

O'Reilly Radar 文章提出 AI 系统存在"保存缺口"：运行时依赖（模型端点、检索数据、策略、工具契约等）在执行时才确定，现有基础设施只保存模型、代码、部署和观测记录，不保存某次执行的实际绑定关系。作者为此提出参考架构 Orrery，用于保存重建 AI 系统历史执行所需的运行时绑定与依赖状态，并区分了"可复现性"与"可重建性"两个概念。

Verifier scores now serve as both benchmark metrics and training rewards for large language model (LLM) agents, and a change in score is routinely read as a change in capability. It may instead reflect a change in the evaluation. We introd…