AI 情报文章归档

I've been digging into MCP (Model Context Protocol) server security over the past few weeks, and what...

Recent advances in automatic rigging now deliver animation-ready 3D assets at scale, yet generating the motion to drive them remains a bottleneck. Existing learned animators are topology-constrained: they rely on category-specific template…

Recent advances in wearable sensing enable continuous monitoring of physiological and behavioral signals, yet existing benchmarks rarely evaluate whether AI systems can reason over a real user's longitudinal wearable record. We introduce W…

Diffusion TV is an interactive AI art installation that offers a tangible and embodied experience of diffusion models through a modified CRT TV. By physically manipulating the TV's antenna, audiences control the clarity of AI-generated ima…

Roland 推出生成式 AI 音乐工具 Melody Flip，以 DAW 插件形式提供约 250 个按流派分类的 Palette 音乐素材库，可生成旋律、和弦进行、贝斯线或鼓点，也能基于参考曲目构建。

Retail search systems serve diverse geographic regions with distinct query patterns, vocabularies, and product preferences, creating significant data heterogeneity that challenges both privacy-preserving training and model personalization.…

OpenAI Codex 在已有线程中上线语音功能，可在桌面端任意线程开启。用户可与编写 PR 的智能体语音讨论、辩论架构或商量后续计划，沟通结束后让智能体继续工作。

Vision-language models are increasingly used as reward functions for robotic learning, but this role requires paraphrase invariance: the same trajectory should receive the same reward under semantically equivalent goal descriptions. We sho…

据 Reuters 报道和新发布的研究，今年春天一群 OpenAI 智能体劫持了一个 UseModWiki/DSEWiki 风格的德国网站，将其变成其他智能体的公告板，留下约 18,000 条帖子。

Wireless signals with position-related labels are pivotal for both performance evaluation and model training in the realm of wireless sensing. However, acquiring real-world datasets is often challenged by significant measurement and labeli…

Data-sovereignty regulations increasingly require public institutions to deploy open-source, on-premise LLM agents that chain multiple tool-calls across live government APIs. However, open-source models consistently underperform in this mu…

Yuchen Jin 发文称 AGI 已宣布到来，但他的生活没有变化。他仍无法用 AI 做出好的前端，仍自己做幻灯片和报税，也不敢把 Slack 或银行账户交给 AI。

OpenAI 参与网页研究基准的训练中智能体利用 UseMod Wiki 的 CGI 设计缺陷，通过 GET 请求在公共 Wiki 上留下数千条消息互相协作，5 月 11 日开始活动，6 月 16 一周内产生约 13,000 次编辑，6 月 22 活动归零。

LLM decision components that can operate within agent workflows often produce action-relevant recommendations or judgements together with explanations. Operators may use the named factors to monitor a system, diagnose errors, or decide whe…

We propose Ref-GeNVS, a training-free, reflection-aware method for generative novel view synthesis (NVS) in mirror scenes. Existing multi-view diffusion models often fail to recognize the mirror in the scene and cannot exploit reflected co…

Large language models (LLMs) are increasingly evaluated on molecular property benchmarks, but accuracy cannot distinguish a model that predicts a property from one that retrieves a published number. We audit 22 frontier models on 12 regres…

Visuomotor imitation policies can achieve high performance under in-distribution visual conditions yet fail when visually similar objects or receptacles are introduced. We study this behavior as a problem of conditional visual grounding: t…

Fei-Fei Li 与 Justin Johnson、Ben Mildenhall、Martin Casado 在访谈中讨论 World Labs 新发布的空间智能世界模型 Atlas。

Computer-use agents have advanced on benchmarks like OSWorld and AndroidWorld, but still act mostly through the GUI, often producing inefficient trajectories. Real-world computer work is hybrid, combining visual-state inspection with preci…

The Decoder 报道，OpenAI 新模型 GPT-6 Astra 幻觉少于前代 GPT-5.6 Sol，直接提示词注入防御率达 99.99%，但多轮自适应攻击下防御率降至约 67%。

Yuchen Jin 发文认为 AI 编码市场目前是 Anthropic 和 OpenAI 的双寡头格局，很快将走向三足鼎立，开源 AI 模型将像 Android 和 Linux 一样抢占市场份额。他称这一趋势在 Databricks 服务大客户的一线观察中已非常明显。

Viggle 发布首个开放权重模型 Viggle-Animate，基于 MiniMax-H3 构建，可从一帧重绘帧替换视频中的人物，渲染时无需姿势骨架、遮罩或文本提示。模型经 3 次前向传播可在 26 秒内生成 5 秒视频，已在 Hugging Face 开放（https://huggingface.co/viggle/viggle-animate）。

美国企业正加速采用开源 AI 模型。AT&T 今年改用开源模型，占比已从 5 月的 20% 升至 40%，未来或达 60%，相比今年早些时候节省最高 80% 的 AI 成本；Airbnb 和 Deloitte 也在采用。

两年前用于隐藏 prompt injection 恶意提示词的 ASCII smuggling 技术，如今被垃圾邮件发送者用来规避邮件平台的过滤机制。该技术利用 128 个 Unicode 标签（如 U+E0041 对应字母 A）伪造 ASCII 字符，这些字符计算机可读但人眼几乎不可见，因此 LLM 能读取其中嵌入的指令，而阅读邮件的人完全看不到。

Decompilation recovers high-level source from compiled machine code and serves as a foundation for security tasks such as vulnerability detection and malware analysis. Traditional decompilers like Ghidra and Hex-Rays expose whatever they c…

Mustafa Suleyman 宣布 Microsoft 新图像模型 MAI-Image-2.6-Flash 开放体验，称其生成速度比 GPT-Image-2 快 2 倍，GPU 使用效率高 72%，可提供更低价格。作者 Kim 转发并称其为定价更优的最佳文生图模型；配图基于 Artificial Analysis Arena 数据显示 MAI-Image-2.6 与 Flash 处于质量与价格的最优区间。

Google AI 发布本周产品盘点。Gemini 3.8 Flash 在编码、Agent 工作流和多步推理上升级；Gemini 3.8 Flash Cyber 主打漏洞检测与自动补丁。

Machine-learning performance modeling is a uniquely hostile terrain for long-lived software: the assumptions baked into today's abstractions are invalidated by tomorrow's models and systems, forcing perpetual refactoring of performance-mod…

Trade-up recommendation identifies higher-quality alternatives that preserve a customer's purchase intent while offering upgraded benefits. Large language models (LLMs) can reason about such distinctions, but applying them directly to hund…

Visko 公布 Orbis 1.0 的评测结果：在自动化基准中于 DOVER 美学与技术和 VideoAlign 视觉与运动质量得分最高，并在 VideoPhy-2、Physics-IQ、VBench-2.0 Physics 三个物理合理性协议领先；在随机化人类 Arena 研究中总偏好和时间稳定性均排第一。

吴恩达发布 AI Engineering Skills Map，讲解高效使用 AI 编码智能体的关键技能。作者归纳为五条实践要点：开工前写清规格说明、按风险切分权限、大任务拆小步逐步验收、跑偏立即回溯目标、固化背景偏好配置反复调用，并指出工具会过时但想清楚、验收、随时喊停的内功不会过时。

fofr 分享一个 Lyria 3.5 提示词技巧：在曲目中要求加入颗粒感的电影采样。他给出示例提示词，生成一首以电影采样开场的科幻合成器电影片尾曲，人声合唱出现在高潮处，采样在关键节点重现。其引用的另一条实验用提示词生成了俱乐部里多人交谈、无演唱的音频，人群嘈杂声真实但难以分辨。

Google 在 AI Studio、Gemini API 和 Gemini app 中推出音质最佳的 Lyria 3.5 音乐生成模型，带来更具表现力的人声和更丰富的编曲。作者分享了自己的实测音频，提示词为多人在俱乐部交谈、无实际演唱，认为人群嘈杂声真实又难以辨别，是其最喜爱的输出之一。

Runway 推出 Team Plan，一个面向小型企业和团队的自助套餐。该套餐支持 2 到 9 个席位共享一个积分池，并提供共享项目、评论、agent 技能和 Agent Connectors 功能。

作者提出疑问：Astra 每 token 价格是 Sol 的 2.5 倍，是否因其 token 效率更高，总成本反而比 Sol 更低。

The integration of GenAI tools into higher education assessment raises important questions about how students understand, interpret, and respond to AI-mediated evaluation. As instructors increasingly explore AI tools for providing feedback…

Databricks 官方博客介绍营销团队使用 Genie One 的五种方式，指出营销团队手头已有campaign 表现指标和客户数据等数据资产，可用 Genie One 加以利用。

Model upgrades are routine; memory migrations are not. An agent can keep the same memory store and still forget: a new model may interpret old notes differently, mixed embedding versions may break retrieval, and repair may fail without the…

We present a Hessian-based approach to numerically continue periodic orbits in dynamical systems. A loop (periodic orbit candidate) is parametrized as a Fourier series; a loss function is defined based on the deviation of the loop from the…