AI 情报文章归档

Built for the All Things Agentic Hackathon. The project: an autonomous compliance agent for cannabis...

Discover how the General Availability of tunable CrashLoopBackOff in GKE speeds up distributed AI/ML workloads and sidecar initialization without host-level hacks.

论文《Teach the Magnitude， Not the Direction》提出 CREST 框架，为多轮多步 LLM 智能体做分层信用分配：每轮单独获得验证的信用，再由同一模型充当自教师，对轮内不确定决策加大学习权重，但教师只能调节更新幅度、不能推翻验证器的判断。

Six failures from the first twenty-four hours of an autonomous agent loop, including the monitor that could not detect the loop stopping, and the book that told its buyer it could not be bought.

Sony、EMI 和 Warner Chappell 等音乐出版商于周五起诉 Anthropic，认为其在图书作者案中支付的 15 亿美元和解金额不足以形成威慑，并指控其通过 BitTorrent 从 LibGen 和 PiLiMi 下载了含歌词和乐谱的书籍。

Grok Build 发布 v1.0.14 更新，主要提升 Grok CLI 的可靠性与工作流。OIDC token 刷新改为默认主动进行，PostToolUse hooks 可在工具执行后向模型反馈信息，grok usage 支持查看每轮持久化的 token 与成本数据，Windows 下载体积缩小约 70%。

英国央行行长兼金融稳定委员会主席 Andrew Bailey 致信 G20 财长，警告 AI 估值虚高叠加杠杆投资上升，一家大型 AI 公司受挫可能拖累其他科技巨头乃至整个市场。他还指出前沿 AI 将改变网络攻击的速度、规模与成本，而许多国家尚无针对高级 AI 模型的规则，呼吁全球推进安全的 AI 模型发布。

OpenAI Developers 宣布 WebMCP Challenge 办公时间答疑已开始，团队与来自 ChromiumDev、CloudflareDev、ShopifyDevs、Vercel、Render 和 Netlify 的支持者一起解答参赛者的 WebMCP 构建问题。活动于 8 月 31 日上午 11 点（PT）在 Discord 举行。

Sierra 宣布 Julia Brau Donnelly 将加入公司，担任首席财务官（CFO）。她将负责公司的财务战略与运营，助力 Sierra 在 AI 客服领域的持续扩张。

While WhisperX accelerates speech transcription via intra-audio batching, it isolates audio segments, losing the historical context needed for coherent punctuation and terminology transcription. Conversely, standard Whisper retains context…

Bridging model-based control and learned policies in long-horizon manipulation has harbored a silent disagreement: control executes specified objectives, learning amortizes that behavior into a reactive policy, yet existing protocols disca…

Uncoupled no-regret dynamics provide a decentralized route to equilibrium, but prior guarantees for individual regret retain a polylogarithmic dependence on the horizon. We remove this dependence for every finite $N$-player normal-form gam…

Many parameter-efficient methods generate the parameters of a large neural network from a low-dimensional latent representation. Given an architecture $Φ$ with $P_Φ$ parameter slots, we write $\boldsymbolθ_f=\mathcal{G}(\boldsymbolξ_f)$, w…

我们如何改进 Grok @bot？你希望看到添加哪些功能？

The 2025--2026 AI market has seen a wave of stealth releases: frontier models launched anonymously on developer platforms under codenames. For their users, identity determines data-handling terms, supply-chain risk, and capability expectat…

BioMedRAG introduced retrieval-augmented generation with a learned chunk scorer for biomedical information extraction. However, it relies on fixed-size chunking which can fragment semantic evidence. We propose a configurable semantic chunk…

Ontology alignment (OA) has evolved through several methodological paradigms, ranging from lexical and structural aligners to knowledge graph embedding (KGE) models and, more recently, Large Language Model (LLM)-based approaches. Although …

Neural network mixed-effects models (NMMs) have gained traction by combining the strong representation and predictive power of artificial neural networks with the capacity of mixed-effects modeling to capture complex correlation structures…

Large language models (LLMs) offer promising clinical decision support but remain vulnerable to hallucinated facts, unsupported recommendations, and citation errors. We present DIASENTINEL, a fully on-premise multi-agent system for one-yea…

a16z 的 David George 在 Gavin Baker 的访谈中表示，西海岸 AI 发展势头迅猛，他试图寻找持看空观点的人，但无人能给出一个正在恶化的量化数据点。他称至少在 7 月和 8 月，找不到一个这样的人。

The motivation for this paper is the investigation of the trade-offs implicit in probabilistic models used in machine learning. Models are often used to make predictions in the form of conditional probabilities. However, a pair of conditio…

Research planning is the decisive capability of AI scientists. Yet a research plan admits no verifiable answer, so reinforcement learning lacks the environment it requires: tasks paired with a critic. Rubrics extracted from scientific pape…

The effect of Large Language Model (LLM) scale on ontology learning (OL) performance remains insufficiently characterized. We present a controlled evaluation of 13 models spanning dense and Mixture-of-Experts variants from the Qwen3.5 and …

如果你不想等待体验 HF 开源机器人的魔力，你应该入手一台 Reachy mini，发货比 microduck 快得多，而且同样有趣！https://store.pollen-robotics.com/

Google 提出 SKILL.state，用显式可变执行状态替代追加式对话历史，解决长程智能体运行变慢和上下文污染问题。模型每步仅读取不可变技能规范、当前结构化状态和最新观察，中间推理在生成有效状态更新后即被丢弃，提示词不再随运行增长。在多个数据集、模型和执行环境中，任务准确率提升，累计 token 消耗下降，且该抽象与架构无关，可移植到现有技能运行时。

WSJ：AI 正迫使 EY 改变其奖励员工的方式。 随着 AI 承担更多日常专业工作，EY 投入 1 亿美元支持人类判断力。 公司表示，员工因适应性、创新、判断力和 AI 实验可获得最高 500 美元的即时奖励，重大个人贡献可获 1 万至 2.5 万美元奖励。 值得注意的是，EY 的 AI 相关收入在 2025 财年增长了 30%。

Instagram 宣布将"AI 创作者"标签改为"AI 生成账号"标签，因用户常无法区分 AI 账号与真人。未标注的账号将限制触达和推荐，受影响账号可通过 Account Status 申诉；仅将 AI 作为工具的创作者无需标注，选择标注者保留完整触达。

Generative models have become central across science and industry, from image and text synthesis to the design of molecules and materials. Quantum generative models are considered one of the most promising applications for quantum computer…

一份 ChatGPT 工作工具与技能参考文档上线，汇总了相关工具与技能的使用说明。该参考面向开发者与进阶用户，便于快速查阅 ChatGPT 工作场景中的功能配置。文档以站点形式发布，供社区参考使用。

有点惊讶，我们并没有看到更多围绕AI能力构建的激进政治理念--就像早期大规模工业化曾是资本主义、共产主义和社会主义思想形成的主要推动力那样。

智能体群集化

Many important forms of human learning begin with a vague goal, such as "become a better physicist" or "improve at research." Learners must interpret the goal, identify capability gaps, decide how to learn, and determine whether they have …

Efficient evaluation changes the protocol used to support claims about model behavior, yet it is rarely tested whether those claims remain stable after the evaluation itself is made cheaper. We stress-test conclusion robustness in responsi…

I built a pixel-art office where ChatGPT, Gemini and Grok work as actual software engineers. Then I gave them access to real repositories.

Users of a deployed language model routinely encounter behaviours that testing almost never surfaces, since deployment puts the model through orders of magnitude more interactions than any evaluation can simulate. Automated auditors make t…

Industrial post-training is a brownfield regime. Teams inherit a deployed checkpoint and must land targeted improvements under fixed compute and mixture budgets without regressing the rest. The maintained artifact is increasingly dataware:…

Google 等机构提出 SKILL.state，用显式可变执行状态替代不断增长的对话历史，解决长程任务中智能体变慢和上下文污染问题。模型每步仅读取技能规范、当前状态和最新观测，中间推理在生成有效状态更新后即被丢弃。在多个数据集、模型和执行环境中，任务准确率提升，累计 token 消耗下降，且该抽象与架构无关，可移植到现有技能运行环境。

Why a machine must be allowed to be wrong I built meta-science for the All Things Agentic...

Grok Bots 将 LLM 引用数据转化为一个运营闭环： 找到 ChatGPT、Grok、Gemini 和 Perplexity 已经引用的页面，联系那些页面的控制者，然后重新运行相同的查询，看看可见度是否发生变化。 AI 搜索正在改变品牌被发现的方式。 你现在可以通过与 MCP 对话来在 ChatGPT 中排名。 这里的思路是将引用本身视为可观测性数据。

Large language models (LLMs) increasingly interact with external environments and accumulate substantial behavioral experience, yet existing agent benchmarks largely evaluate them as fixed policies. It therefore remains unclear whether an …