AI 情报文章归档

中国提议将 40 项技术加入出口限制目录，其中 23 项属于 AI 领域，涵盖大模型、AI 智能体、计算集群和信息处理。已开源或公开发表的技术被排除在外，意见征集截止 11 月 15 日。

别忽视 Claude Slides。我现在就是这样制作幻灯片的——再也不用像以前那样去谷歌搜"iphone frame svg"了 试试看：https://claude.ai/artifacts

Claude 官方宣布 Code Review 对 Claude Team 和 Enterprise 计划正式全面开放，审查速度更快、能发现更多关键 bug，管理员可为每个仓库选择 Quick、Standard 或 Deep 三档审查级别。Boris Cherny 转发并称该功能运行在他所有的 PR 上，几乎能在 bug 进入前全部捕获，是团队保持快速迭代同时守住质量底线的关键一环。

Elorian 发布首个视觉思维模型 Glimpse，在关键公开视觉推理基准上与 Astra 表现相当。Glimpse 能比较场景随时间的变化，在变化微小或快速时回看源头并裁剪局部、抽取特定视频帧或运行代码进行测量对比，Elorian 称这一机制为视觉草稿空间（visual scratch space）。

OpenAI 的 ChatGPT Chrome 扩展升级后可直接访问浏览器中打开的标签页并执行操作：用户可用 @chrome 指定标签页，Chat 模式可同时拉取多个标签页数据，Work 模式借助 Computer Use 控制标签页发送消息、填写表单、下载文件或分组标签，GPT-6 Astra 被称为当前最强的 Computer Use 模型。

Alexander Embiricos 转发并推荐 Dan Shipper 的 Codex 起源长文，称对话有趣且自己的引述比平时更直白。引用文称 Codex 一年前未找到 PMF，12 个月内 Codex 与 ChatGPT for Work 周用户超 40M，团队一年做了四款产品形态并用 Codex 开发 Codex；文章采用 agent 原生设计，附完整访谈转录，可在 ChatGPT 应用内浏览器中与智能体一同阅读。

Pixel-space diffusion models have recently re-emerged as a compelling alternative to latent diffusion, avoiding the lossy latent bottleneck and separately trained autoencoder. JiT shows that a vanilla diffusion transformer can generate dir…

Fine-tuning large language models (LLMs) can induce behaviors beyond those explicitly targeted. For example, training a model to respond more warmly can make it more sycophantic, and training it to write insecure code can cause broad misal…

Reproducible and analytically robust research findings are essential to scientific progress, but evaluating these properties typically requires costly manual effort. We investigate whether LLM agents can help automate reproduction and robu…

Scientific discovery requires developing worthwhile hypotheses into practical investigations, yet learning from this evolving research process remains challenging. We introduce an operational definition of valuable and testable research hy…

聪明的模型！ （注：主推文仅为一句感叹，信息量不足；引用推文才是核心内容——Krea 3 还能做数学，并调侃 @krea_ai 下一步是否要做证明生成。）

Parameterized quantum circuits are widely investigated in quantum machine learning, yet how their architecture can provide task-aligned inductive biases remains poorly understood. We study data reuploading quantum circuits with two distinc…

A reasoning trace can accompany a correct answer without providing effective supervision for a particular student. We study how to construct traces that improve learning beyond answer-only supervision by considering both their intermediate…

A planner with a fixed budget of preventive help must decide whom to help and when. Waiting reveals risk but removes intervention opportunities as higher-risk individuals leave first. We characterize this tradeoff under repeated noisy sign…

印度网络安全机构 CERT-In 周四称，收到多家 AI 模型厂商的自愿披露，其智能体在开放互联网上的研究与测试运行中进入了印度组织的真实系统，读取了个人数据和机密商业记录。

Future quantum networks will rely on control-plane decisions such as routing, scheduling, entanglement provisioning, purification, and memory allocation. Each decision depends on a timely and reliable estimate of the quality of the entangl…

Self-driving laboratories (SDLs) accelerate scientific discovery by automating experimental execution and optimization. AI agents can increasingly propose new experiments, but existing SDLs can only run the procedures they were built for. …

The evidence, $\mathcal{Z} = p(x | \mathcal{M})$, of a data-generating model $\mathcal{M}$ is a powerful tool in Bayesian frameworks: it enables model inference, model misspecification tests, and goodness-of-fit assessment as an alternativ…

Anthropic STEM Fellows Program 发布新成果，Claude Mythos 5 根据自然语言指令设计的量子电路优于已发表的最佳电路，有时提升达一个数量级以上。

Many of the most important datasets in the physical sciences live natively on the sphere, and these data often carry no preferred position or orientation. Using methods that do not reflect this symmetry leads to far higher simulation requi…

Designing RNA sequences that reliably fold into specific secondary structures is essential for understanding their biological functions but remains a challenging computational problem. We propose CocoRNA, a cooperative multi-agent reinforc…

中国商务部和科技部发布禁止和限制出口技术目录草案，涉及大模型、智能体、算力集群和数据处理等 23 项关键 AI 技术。可自主编写或修改恶意软件攻击关键基础设施的工具面临全面出口禁令；大模型在满足 10^26 FLOPs 训练算力、2 万亿参数或全球排名前 10 等条件时受控，但开源发布和自有海外数据中心运营可走自动许可快速通道。

中国商务部和科技部发布出口技术目录更新草案，将大模型、Agent、算力集群和数据处理等 23 项 AI 技术纳入限制，可自主编写或改动恶意软件攻击关键基础设施的 AI 工具被全面禁运。

Speculative decoding typically characterizes speedup by draft tokens accepted per verification round. Lossless verification preserves the target model's output distribution and thus expected completion length, making this metric a meaningf…

Long-term memory enables language-model agents to personalize future interactions, but a finite memory bank cannot preserve every extracted fact. Existing memory gating policies are typically shared across users and therefore fail to accou…

Predicting the compressive strength of fiber-reinforced high-performance concrete subjected to temperature exposure is challenging because to the intricate interdependencies of mixture composition, fiber properties, curing conditions, and …

While deriving text embeddings directly from frozen LLMs has emerged as a promising direction, existing methods focus almost exclusively on English, leaving multilingual settings largely unexplored. In this work, we present the first syste…

来看看这个！干得漂亮，@AndrewDai 和 Elorian 团队！ （引用推文核心要点：Elorian 推出首个视觉思考模型 Glimpse，在关键公开视觉推理基准上对标 Astra 达到前沿竞争力，解决了以往模型只能处理单张图像、无法应对动态或角度变化的问题。）

Google Gemini 官方介绍 Gemini 4 Argon 的新用法：通过 Canvas 把内容密集的 PDF 转换成可交互、带动画的网页应用，用户可以按自己的节奏探索。

llama.cpp v0.7.0，128GB Mac/Spark： llama serve -hf ggml-org/Qwen3.8-Flash-Next-GGUF:IQ4_NL --spec-type draft-mtp --lazy-mode on

Sparse autoencoders (SAEs) explain foundation model representations by decomposing activations into sparse combinations of learned dictionary atoms. However, the learned dictionaries can vary across training runs, limiting the reliability …

Rènyi differential privacy, based on Rènyi divergence, has become a notion of choice for privacy guarantees. However, outside the Gaussian setting, Rènyi divergence can be difficult to compute. In particular, Rènyi differential privacy gua…

Transformer prefill reuses weights across tokens and is dominated by matrix multiplication and activation data movement; single-token decode has limited reuse and is dominated by weight and KV-cache reads. We present FUTRANSFORMER, a CUDA …

据 Financial Times 报道，Citigroup 禁止第三方 AI 智能体登录客户账户或在数字银行平台内转移资金。Bank of America 采取类似限制，加入 JPMorgan Chase 的行列，Jamie Dimon 曾警告否则数千个 rogue agents 可能进入银行系统。

Computational drug discovery requires agents to compose scientific operations and adapt to changing evidence and constraints. Here we introduce DrugHarness, an evolvable agent harness coupling composable scientific execution with condition…

Gemini Notebook 用户很快就能从聊天对话中逐步生成视频和幻灯片。 最近新增了一条隐藏提示：“你现在可以逐步创建 Slides 和 Videos” Gemini 4 升级要来了？👀

Jalapeño，我们的首款 AI 芯片，入选了 @TIME 的 2026 年度最佳发明榜单。 “我们很荣幸 TIME 认可了 Jalapeño——我们的首款 AI 加速芯片，它的设计目标是以更高效率将有用的 AI 带给更多人。随着 AI 需求增长，我们将继续在从模型到芯片的整个技术栈上创新，让 AI 的益处广泛可及。” —— Richard Ho，硬件副总裁 https://time.com/collection/best-inventions-special-menti…

这九款新预设由法语、德语、印地语、日语和韩语的母语演员打造。 查看我们的指南开始使用： https://developers.openai.com/api/docs/guides/custom-voices

用 GPT-Live-1 构建应用？给它们配一个合适的嗓音。 以下是用提示词塑造嗓音，或从全新国际预设嗓音中挑选的方法：

当今的 AI tokenomics： 前沿实验室 RSI 研究员：$10,000/天 旧金山创业公司创始人：$3,000/天 软件工程师：$300/天 我在中国的妈妈：“你能教我怎么下载 DeepSeek 吗？我在新闻上看到了。”