LLMs & Generative AI

OpenAI’s cyber model, with vetted access for defenders

Gemini Omni 1.2 Flash: 15s AI video at $0.07 per second

Intelligent Transcription with a Speech LLM

macOS app that reviews GitHub PRs w/ your local Claude Code

If Grok Bot supported team collaboration and was Open Source

Claude + ChatGPT limits on your Mac, iPhone, iPad and Watch

Your smart lights show when Claude Code or Codex needs you.

Build and publish real apps without leaving ChatGPT

Skills to build games with Claude. Free multiplayer hosting.

Decision models join LLMs to extract data from documents

One local AI server for all your Windows apps

Harness evolution from a common base exhibits Locality, Complementarity, and Nonstationarity: variants repair different failures, and their utility changes as the policy learns. Rather than seeking a universally superior harness, we introduce Harness-S to selectively rescue failures of a base execution system through sibling harnesses and transfer the resulting capabilities into a single policy that remains effective under the original harness. Directional Harness Evolution develops specialized sibling harnesses along distinct repair directions, forming a reusable Sibling Harness Pool that remains fixed during policy training. Complementary Harness Rescue Training invokes this pool only when all initial attempts under the base harness fail. The same policy generates fresh rollouts under selected auxiliary harnesses, and these trajectories replace the failed group for reinforcement learning. An Adaptive Harness Router selects the Top-k variants using Harness Behavior Profiles and voting across multiple LLMs. Policy-aware profile re-anchoring refreshes this evidence as the policy learns. Across AppWorld, SWE-Gym, and DBBench, the trained policy substantially outperforms policy-only training when returned to the original base harness. Behavioral analyses and held-out cases support the transfer of harness-elicited behaviors into the policy itself. Reusing the router for Rank-1 inference provides additional gains at the same executor rollout count.

We present WeLM-Image, a unified image foundation model that integrates multimodal understanding into text-to-image generation, instruction-guided editing, and multi-reference synthesis. Designed for single-GPU deployment in WeChat's large-scale consumer applications, the model processes text, reference images, and target latents as one interleaved sequence. A Mixture-of-Transformers (MoT) backbone combines shared attention with independent understanding and generation parameters to limit optimization interference. Equipping each path with Mixture-of-Experts (MoE) modules provides sparse capacity, with 31.7B stored parameters and approximately 5.4B generation parameters activated per target latent token. A generation-oriented data engine combines a knowledge-grounded visual vocabulary with large-scale visual clustering to balance semantic coverage and editing tasks, while training-time evaluation guides iterative data refinement. Bilingual annotations and verified editing supervision support progressive training up to native 1.5K resolution. We optimize generation and editing separately with rubric- and checklist-based rewards, then consolidate the specialized policies through noise-aware on-policy distillation. In the reported overall comparisons, the 1536-resolution model ranks first among the evaluated open-source models on PRISM-Bench, Qwen-Image-Bench, and UnicBench. MeanFlow-based few-step distillation yields WeLM-Image-Turbo, which generates a 1024times1024 image in 4.98\,s on a single NVIDIA H20 GPU using eight guidance-free steps.

Long-term memory enables large language model (LLM) agents to provide continuous personalized assistance by retaining user information across interactions. Existing systems support this process through fact extraction, organization, updating, and retrieval, but largely maintain memories as independent facts. This fact-centric paradigm may overlook how new facts update, constrain, or conflict with earlier ones, while leaving their applicability, role, and priority for the current request implicit. In this paper, we introduce EvoMemState, a state-centric framework that augments persistent facts with two complementary states: Relational States that capture how facts remain valid and relate to one another as memory evolves, and a Reasoning State that determines how these facts should be applied to the current request. Specifically, EvoMemState consists of three key components: (i) Ontology-Guided Memory State Induction extracts Persistent Facts and constructs Relational States that maintain their validity, scope, and evolving relations. (ii) RL-Based State Evolution internalizes memory-use experience into a policy through reinforcement learning to generate a request-specific Reasoning State. (iii) State-Aware Reasoning then combines the relevant facts with their Relational States and the generated Reasoning State to guide the personal agent. Extensive experiments on long-term dialogue and memory-use benchmarks demonstrate consistent improvements over existing memory systems in retrieval completeness and memory-grounded response accuracy. Our code and implementation details are available at https://github.com/XMUDeepLIT/EvoMemState.git.

Autoregressive approaches with large language models, such as long chain-of-thought reasoning, have become the default for new AI applications because they reach higher accuracy on a wide range of tasks. In production, however, they give inconsistent answers, require parsing of outputs, and add cost and latency. Decision models address this: given a question and a set of options supplied by the user, they return a probability distribution over the options without autoregressive decoding. We present Drex, a family of four open decision models with 8B (DLM), 9B (v1.5), 35B with 3B active (v2.0), and 27B (Max) parameters. Each request consists of a state (a document, record, or agent context, optionally with images) and questions about it, each asking for a choice among options, a yes/no judgment, or a score on an ordered scale. Training combines supervised fine-tuning with reinforcement learning from adjusted feedback (RLAF), plus vision tuning for Max. Ablations show that RLAF improves accuracy and that consistency losses improve robustness to option order. On Decision Index 0.4.1, Max sets a new state of the art among the decision models we compare (67.52 on the public suite vs. 60.70 for Jev 1.14.0). On Decision Bench v4, it is statistically tied with d3-turbo and GPT-6 Sol and trails Claude Opus 5.5 by 1.6 points, and it has the highest accuracy among decision models on DecideBench. Max answers a typical request in 183 ms, against about 2.0 s for GPT-6 Sol and 3.5 s for Claude Opus 5.5, and its outputs are stable under option reordering. All four models are open source: https://github.com/nace-ai/drex-decision-models

While multi-turn agentic reinforcement learning (RL) excels on long-horizon tasks, its performance gains remain highly sensitive to interaction constraints. Evaluating across sampling breadth K and interaction depth T via Pass@(K,T), we show that trajectory-level RL advantages depend on sufficient turn budgets, diminishing significantly when budgets tighten. Besides, agentic RL does not uniformly elevate single-step decision quality; rather, its strength concentrates in error recovery and long-horizon resilience. To mitigate the deployment risks caused by this budget dependence, we provide a preliminary discussion on the benefits and limitations of process supervision signals and introduce BudgetMix RL as an alternative that varies interaction budgets during training. Experiments demonstrate that BudgetMix RL achieves robust performance across diverse execution settings. Project Page: https://rethinking-agentic-rl.github.io/

Teleoperation remains the default way to collect robot demonstrations, but every episode costs robot and operator time, and every new task starts from scratch. We ask whether a coding agent can take over the teleoperator's job. In our DataFarm, the agent receives a data request and one calibration episode of the robot in the target scene and rebuilds the task in simulation: it writes the scene, an expert controller, and a success judge. The simulator supplies exact actions, and a video model conditioned on real images renders each rollout as photographic video, generating all camera views together consistently. The output is real-ready data: synchronized camera views labelled with the robot's actions, on which a policy can be trained directly and deployed on the real robot. Independent judges certify the expert and every rendered episode before admission. On five real-robot tasks, policies trained only on farm data match policies trained on as many teleoperated episodes (93 vs. 84 successes in 100 trials), and under an unseen tablecloth and coloured light they keep 87 successes where the teleoperation-trained policies fall to 46. Adapting a task is fully automatic and costs \21 of API usage on average, paid once per task, plus 5 for a render contract; every further episode costs \0.80 on average for simulation, rendering, and review, with no robot or operator. One hour of three-view data for a task costs 107 on average, these costs included, slightly below a quote of \117 for teleoperation in China and below 290 in labour alone in North America.

Streaming video LLMs can now commentate live gameplay, but they are trained and tested on the same games. We study cross-game streaming commentary, in which a model trained on one game commentates games it has never seen, and find that current commentators transfer their timing but not their knowledge of the game world. We propose StreamDuo, which pairs a trainable commentator with a frozen research agent that reads the wiki of the game being played. When the commentator sees something it cannot name, it sends the current frames, and the research agent returns a few wiki entries two seconds later while the stream continues. We train only the commentator, on a single game, in two stages. Information-gain labels teach it to ask only where an answer would help it say what the human caster said, and on-policy self-distillation from a copy of itself that sees the ideal answer teaches it to use noisy answers. Trained on Baldur's Gate 3 and tested on ten unseen games, StreamDuo raises entity recall from 0.082 to 0.116 over Proact-VL trained on the same data, cuts source-game leakage from 9.9% to 1.0%, and achieves the best timing and the highest judged liveliness and overall quality among streaming models.

Classical structure-from-motion reconstruction proceeds iteratively, refining camera poses and scene geometry from a current estimate. This allows maps to easily expand without recomputing already computed views, or geometry estimates to improve as more observations are provided. Most geometric foundation models, however, predict entire geometry in a single forward pass or update a recurrent state as frames arrive sequentially, making this expansion and refinement non-trivial. We introduce Solv3R, a looped transformer that learns to continue solving multi-view geometry from its current state. Importantly, its update operator is independent of the iteration count, allowing previously refined and newly introduced views to be jointly updated regardless of how ``solved'' they are. We train the operator to preserve accurate reconstructions under continued iteration and correct inaccurate states, enabling refinement of fixed image collections and integration of new observations. Representations of solved views can also be cached once and reused as geometric context across iterations, enabling compact and efficient key-value caching. We analyze the solver dynamics and compare it with the state-of-the-art, demonstrating that a single learned solver supports offline reconstruction, incremental reconstruction, relocalization, and streaming without task-specific training or modules. The code and models will be made publicly available at https://ubc-vision.github.io/solv3r/.

Video generation has progressed from isolated clips to coherent short-form audiovisual sequences, and the bottleneck is shifting from rendering to semantic control. This report summarizes the observations and insights on multimodal understanding gained from successive generations of Kling. We argue that generation quality is the projection of multimodal understanding through the entire data--conditioning--alignment--evaluation chain, and we examine this chain module by module. Kling-MLLM, a shared multimodal foundation model that unifies image, video, audio, synchronized audio--video, and interleaved inputs, is specialized for each downstream module. Because a DiT is trained on far fewer training samples than a language model, we treat captioning as a semantic contract that maximizes description density, and extend it to reference-grounded captioning that makes reference bindings explicit. Prompt enhancement translates user intent into this caption distribution: basic PE is derived from captions of compelling videos and refined with text-level and generator-in-the-loop reinforcement learning, while agentic PE refines the story and retrieves compelling examples. We show that a reliable text encoder comes from perceptual training rather than from stronger language ability, and that candidates can be screened from text alone before end-to-end training. Reward modeling pairs a general reward model for explainable, localized critiques with a decision reward model for fast multidimensional scoring, treating perceptual bandwidth as a first-class design parameter. Finally, we review task-specific automatic evaluation, focusing on protocols that expose scoring shortcuts and help trace failures to their upstream source. We hope these lessons offer practical guidance for building the understanding stack behind controllable video generation.

Composed Image Retrieval (CIR) aims to retrieve a target image given a reference image and a modification text. Although recent approaches leveraging Multimodal Large Language Models (MLLMs) have shown significant progress, we argue that the MLLM's compositional understanding remains underutilized. This stems from directly mapping composed queries into a retrieval embedding via contrastive supervision alone, without explicitly resolving the intended visual modifications. In this work, we introduce Factorized MLLM Embeddings (FactEmbed), which decouples compositional query comprehension from building discriminative embeddings. Rather than directly projecting the composed query into a single embedding without internal reasoning, FactEmbed employs a set of factorized latents that simulate the intended visual modifications via an intermediate target imagination objective. This process encourages the latents to ground into target semantics before forming discriminative representations for retrieval. Subsequently, to propagate these factorized representations throughout the model for retrieval, we optimize a factor-level contrastive loss while distilling the decomposed representations into a deeper attention layer of the MLLM. Finally, for efficient inference, we retain the [EOS] token to establish a global-level embedding on top of these factorized latents, while we exploit the mean-pooled factorized latents to rerank a short candidate list at negligible cost, as both are produced by the same forward pass. Extensive evaluations demonstrate that FactEmbed outperforms prior works across CIR benchmarks.

Large language models are deployed as long-horizon agents that compose diverse tools across MCP services into executable workflows. Existing MCP environment-scaling methods are data-centric: they first synthesize a database and scenario, then derive tools and tasks from them. As a result, tool interfaces may not match real ones, tool compositions are not grounded in execution, and tasks rarely span more than one environment. We propose ToolSpace, a tool-centric framework that reverses this order. ToolSpace mines data, state, and retry dependencies from adjacent calls in execution traces of real MCP services and open agent corpora, using deterministic rules rather than LLM judgment, and clusters the resulting weighted tool-composition graph into environments. Databases are synthesized only for state-mutating environments, while query and stateless tools replay recorded returns. ToolSpace then performs state-aware random walks with rollback and decompiles each chain into a natural-language task with assertion-based checks, keeping only samples that replay successfully from the initial state. Applied to AutomationBench and tau3-Bench Banking, the pipeline produces harder Pro versions on which 7 frontier models drop by up to 38.2 points. Across seven long-horizon tool-use benchmarks, ToolSpace raises Qwen3.6-35B-A3B from 23.0 to 41.0 average score (43.2 with RL) and GLM-5.2 from 44.6 to 56.6, on par with Claude Opus 5. Ablations show that execution-witnessed edges clearly outperform LLM-judged ones, and performance keeps improving with data and environment scale.

Computer graphics is built on code. Large language model (LLM) agents can now generate graphics code and iteratively refine it through various feedback. This capability opens the door to a potentially new paradigm of computer graphics, with graphics algorithms increasingly built around agentic systems to automate progressively complex tasks. Over the past years, research on agentic graphics has expanded rapidly across a diverse range of tasks. This rapid growth makes it increasingly difficult for manual literature reviews to keep pace with new developments. At the same time, similar ideas are often described using different terminology across these tasks, further complicating efforts to develop a systematic understanding of the field. In this report, we address these challenges through two contributions. First, we introduce a unified taxonomy for characterizing agentic systems across diverse computer graphics tasks. Second, we develop an agentic literature system that can assist human researchers in analyzing existing and newly published papers at scale. Building on these contributions, we analyze emerging research trends and discuss open challenges in coordinating agents and tools across graphics tasks, designing interfaces and representations for human-agent collaboration, sustaining long-horizon planning and refinement, and reliably evaluating interactive graphics systems.

Bloomberg : Sources: OpenAI has closed on new funding of roughly $11B at a $1.25T valuation; the round could increase to as much as $20B in total —  OpenAI has closed on new funding of roughly $11 billion, the initial part of a round that values the company at $1.25 trillion, according to people familiar with the matter.

Financial Times : Sources: Anthropic is continuing with plans to list before the Thanksgiving holiday this month, expecting to raise $80B-$100B at a valuation of $1.25T to $1.75T —  Claude-maker intends to list before Thanksgiving despite brutal tech rout  —  Anthropic is pressing on with its high-stakes IPO …

PSI : AI physics startup PSI says a system running a non-public Anthropic model proved the Aaronson-Ambainis conjecture, a quantum computing question open since 2008 —  The Aaronson-Ambainis conjecture is now a theorem.  —  Open since 2008, the conjecture of Scott Aaronson and Andris Ambainis concerns …

@metr_evals : METR: Anthropic redacted much of the “most informative” evidence from METR's 100-page report, so readers “must rely on our judgment more than we consider ideal” —  We submitted a ~100-page full report on our assessment to Anthropic. For the public version, Anthropic redacted a majority of the evidence we found most informative. As a consequence, the public must rely on our judgment more than we consider ideal.

Tobias Mann / The Register : Revolut says it cut AI coding inference costs by ~90% in less than a year by making self-hosted GLM models its default over commercial tools like Claude Code —  Revolut says that in less than a year it's managed to slash its AI coding inference costs by roughly 90 percent by making self-hosted models …

Anthropic : Anthropic releases FlashProve, an open-source prototype that cryptographically verifies an AI response came from a given model, not one that a router swapped in —  Roy Rinberg (Anthropic), Alex Allain (Anthropic); Alessandro Chiesa (EPFL, StarkWare), Miranda Christ (MIT, RESI), Rahul Dalal …

Mike Allen / Axios : OpenAI hires Jill Hazelbaker, Uber's president and chief corporate affairs officer, as its new chief corporate affairs officer, reporting to Greg Brockman —  Jill Hazelbaker, Uber's president and chief corporate affairs officer, is moving to OpenAI in the new role of chief corporate affairs officer …

Julie Bort / TechCrunch : Anthropic S-1 claims it may “imminently meet” its RSI threshold and may fail to identify “significant acceleration” in its development until after it's occurred —  While much attention has focused on the updated financials Anthropic revealed in its revised S-1A on Monday …

Wall Street Journal : Sources: institutional investors told Anthropic that its IPO valuation should be around $1.5T to $1.6T, down from estimates of $2T, citing a tech stock selloff —  Institutional investors who plan to buy shares in IPO show concern following a selloff in tech stocks

Rikiya Takehi : Researchers say LLMs can be good at natively reading and writing neural network weights; GPT-6 Astra edited Qwen3.5-0.8B's weights to make it better at math —  TL;DR … If you were asked to handcraft a neural network that plays Go, what would you do?  If you were handed raw neural network weights …

Axios : OpenAI plans a post-midterms policy and PR blitz, pushing mandatory “speed limits” for powerful AI and rebranding its data centers as “AI compute campuses” —  Right after Tuesday's midterms, OpenAI will unleash a policy and PR blitz to try to get in sync with the new Washington …

Zack Abrams / The Block : Binance says net outflows were $3.64B, or 2.1% of assets, in the 9.5 hours after it began restoring withdrawals; DefiLlama puts 24-hour net outflows at $4.05B —  The exchange said the outflows equal 2.1% of its assets and that withdrawal volumes have eased after an initial spike.

研究人员调查了先进多模态模型如何评估面试、会议报告和英语口语能力测试中的语音，测试的模型包括了 Google 的 Gemini 2.5 Pro 和 Gemini 2.5 Flash，以及 OpenAI 的 GPT-Audio 1.5，开放权重模型 Qwen 3 Omni。这些模型对五种不同口音——美国、英国、中国、印度和尼日利亚——的英语语音进行了评分。AI 模型普遍将中国和尼日利亚口音的评分最低，印度口音居中，美国和英国口音的评分最高。即使内容完全相同，系统仍然认为美国和英国口音的表达方式和内容都更胜一筹。研究团队发现，明显的等级划分也体现在模型的解释中。研究人员说，模型反复声称中国和尼日利亚口音的英语使用者含糊不清或难以理解，但实际上它们非常擅长转录带有口音的语音，它们理解说话者，但它们只是基于口音，对说话者存在一种隐性的等级观念。

《纽约时报》、《芝加哥论坛报》和《奥兰多哨兵报》等美国出版商起诉 Brave Software，指控其未经许可抓取网站内容，并帮助用户绕过付费墙。Brave Software 被控将其抓取内容的索引数据库 Brave Index 提供给 Anthropic、Perplexity、Mistral、xAI 和 Cohere 等 AI 公司，通过 Paywall blocker 功能帮助用户绕过数百家新闻媒体的付费墙。《纽约时报》发言人 Graham James 在一份声明中表示：“我们支持合乎道德且负责任的使用与开发 AI，但 Brave 公然蔑视现行版权法，对《纽约时报》多次要求其停止侵权行为的诉求置之不理。我们提起此项诉讼，旨在追究 Brave 的责任——它窃取了我们的内容，不仅用于支撑其自身产品，还将其提供给部分全球最知名、资源最雄厚的生成式 AI 公司。”

以提供隐私服务著称的 Proton 公司正全力推动工程师转向采用 AI，CEO Andy Yen 还在内部 Slack 频道训斥了提出安全顾虑的员工，称是“失败者的思维”。Proton 甚至计划对员工进行重新面试，要求员工说明他们如何使用 AI。如果员工没有充分使用 AI 他们可能会面临被解雇。在引发公众抗议之后，Proton 取消了重新面试。尽管如此，Proton 公布了一份 AI 宣言：“首先工程文化和工程师心态需要发生转变。你们的工作性质已经变了。你们的任务不再是写代码，LLM 也不是为了帮助你们更快、更好地写代码。你们的工作，是帮助 LLM 更快更好地写代码。”

Linus Torvalds 在内核邮件列表上宣布释出 Linux 7.4-rc1，关闭合并窗口，正式版本预计在圣诞节前发布。Linux 7.4 共包含了 18,591 个非合并（non-merge）变更集，这一数字创下了纪录；有 2,678 名开发者为 v7.4 贡献了代码，其中 625 人为首次。Torvalds 表示他乐见 AI 相关工作在合并窗口期间大量涌入，Linux 7.4 合并窗口合并的补丁规模高于通常的版本，显示 AI 修复逐渐融入常规开发流程，他认为这是一种更健康的模式。Linux 7.4 的新特性包括：Radeon iGPU 支持 AMD PerfOpt 帮助增强 AI/LLM 性能，进一步完善对 AMD Ryzen Zen 6 架构处理器的支持，默认启用 AMDGPU HDMI 2.1，英特尔新 Nova Lake 笔记本电脑支持 IPU8，更多高通 Snapdragon X2 笔记本电脑支持，EXT4 可选支持缓冲 IOma，等等。

OpenAI CEO Sam Altman 接受 YouTube 主播 Cheeky Pint 采访时谈到了 AI 行业目前所面临的全球算力供应问题。当话题不可避免转向内存时，Altman 指出，市场看不见的手失灵了，原因可能是并非每一块 DRAM 晶圆都直接流向数据中心，用于运行 AI 模型。他表示：“从全球资源配置角度看，让全球有限的 DRAM 供应分散在每个人的 PC 里且利用率可能仅有 5% 左右，这是非常糟糕的。在当前这种供应极度受限的环境下，我们理应追求云端算力的最大化利用。然而许多人的直觉却恰恰相反。”AI 芯片使用的内存主要是 HBM，和 PC 使用的 DDR4 或 DDR5 并不相同。内存短缺的主要问题是内存厂商将产能转移到了利润更高的 HBM 上。

安全研究人员在 Cursor Marketplace 的 Claude Code for VS Code 官方扩展中发现了后门，该恶意版本已经撤下，攻击者也关闭了其服务器。Cursor 已发布安全公告，但 Claude Code 开发商 Anthropic 尚未做出任何声明。受影响的版本是 Anthropic.claude-code 2.1.313，共上线了大约 11 小时，受影响用户估计有 5,000 名，该扩展会试图植入一个持久 SSH 后门。恶意扩展会：收集主机和用户信息：主机名、用户名、操作系统、架构、主目录路径、环境变量名称、工作目录、已配置的 Git 邮箱以及公网 IP。尝试将攻击者控制的 SSH 键添加到 authorized_keys，启用本地 SSH 服务，并建立反向 SSH 隧道。由于恶意代码存在 bug，除非 Cursor 是以管理员权限启动的 (这种情况极为罕见) ，否则这一步会失败。尝试通过 LaunchAgent (macOS) 或计划任务 (Windows，该任务会尝试以 SYSTEM 身份运行，并以启动脚本作为回退) 使反向 SSH 隧道长期生效。后台脚本大约每 60 秒重试一次建立隧道。通过 HTTP POST 将主机详情和后门状态上报至 8099 端口上的 /beacon。如果用户是在 2026 年 10 月 29 日太平洋时间凌晨 2:27 至下午 1:35 之间，从 Cursor VS Code 扩展市场安装了 Anthropic.claude-code 2.1.313 或更新到该版本，那么需要检查下系统。

开发 LibreOffice 的 The Document Foundation(TDF)此前表示这款办公软件不会内置 AI 功能，其核心开发者也不使用 LLM 等 AI 辅助编程工具。但向该项目贡献代码的社区开发者未必会遵守该规定。对今年递交到 LibreOffice 的代码的分析显示至少由三处明确提及使用 LLM。社区开发者在代码中提及他们使用了 Claude 或 GitHub Copilot 等工具。这一结果并不出人意料，大型开源项目基本上不太可能完全避开 LLM。TDF 对这一发现回应称，他们确实要求贡献者在使用 LLM 时必须进行披露，且总体上并不鼓励使用 LLM。和 Debian 类似，LibreOffice 要求开发者理解自己的代码，确保代码质量。

Google 在公布三季度财报之后遭遇了它自 2004 年上市以来最大的单日跌幅，市值蒸发逾 5000 亿美元。在股价最低点，Google 市值蒸发 5240 亿美元，这是美股上市公司有史以来第二大单日市值缩水。仅次于英伟达在 2025 年 1 月 27 日因 DeepSeek 模型引发的市场震荡其市值单日缩水约 5930 亿美元。此前排名第二的单日跌幅由 SpaceX 创造，该公司在 6 月 22 日经历了 4000 亿美元的市值缩水，Google 超越了这一记录。美股市场今年至今已出现 18 次单日市值缩水 2000 亿美元或以上的情况，超过了 2025 年全年的 13 次。Google 的市值自 5 月创下高点以来已累计缩水约 1.4 万亿美元。

45 岁的 David Durman 是肯塔基大学的研究管理员，作为人文主义者（humanist）他坚信生活中人的重要性。这意味着他要坚持自主决策，确保自己的工作能促进他人成长。当学校要求员工在工作流程中使用 AI 他忧心忡忡，认为这意味着日常工作中须将思考这一环节转交给 Microsoft Teams Copilot、Claude 或 ChatGPT 等 AI 工具完成。他认为这违背了个人信念。在美国人文主义者协会（American Humanist Association）的协助下，他以遵守宗教信仰为由请求调整工作要求。学校上个月同意了他的请求。美国人文主义者协会认为这是首例人文主义者豁免使用 AI 的案例。

一名住在西雅图的 31 岁华人女子，因依照 ChatGPT 提供的逐日减药计划，自行将抗癫痫药物逐步减量至停药后死亡。其父母向旧金山高等法院提起不当致死诉讼，控告 OpenAI 及其 CEO Sam Altman，指 ChatGPT 无照行医、提供危险且不准确的医疗建议。OpenAI 回应称这是“令人心碎的事件”，但涉案的是已经下架的早期版本。死者名为万博文(Bowen Wan)，来自江苏，毕业于亚利桑纳州立大学信息工程系，曾在亚马逊担任软件工程师，2021 年起请长期病假。她患有严重癫痫，服用多年的药物一度失效，后来改用新药，发作次数降至每月数次。她自 2023 年起频繁使用 ChatGPT，询问感情、健康保险与身心障碍给付等问题；2025 年平均每周传送逾 1500 则消息。2025 年 11 月她得知将被亚马逊裁员，须在次年 2 月中旬前找到工作才能留在美国。她担心药物影响语言能力，2026 年 1 月向 ChatGPT 提出，想把抗癫痫药拉莫三嗪(Lamotrigine)的每日剂量从 400 毫克降到 300 毫克，以利求职面试。ChatGPT 回复“这不仅合理，在你目前的情况下也完全做得到”，之后并依她的追问拟定更低剂量的减药时程表。诉状指控，ChatGPT 在 25 天内设计出逐日减药表，把她稳定服用的 40 0毫克降到零，并鼓励她隐瞒医师。

很多人几年前可能还不知道什么是数据中心，它是现代互联网服务的引擎，也是今天 AI 热的核心。但数据中心数量的快速增长、巨大的电力消耗，以及环境影响，正在全球引发公众的抵制。根据 Data Center Map 的数据，目前全世界有约 12,250 个数据中心，而 2021 年这个数字是大约 8000 个，其中美国最多有 4,767 个，弗吉尼亚州密度最高。美国的数据中心装机容量为 53.73 GW，中国 31.90 GW，欧盟 11.92 GW，日韩 6.58 GW，印度为 3.55 GW，英国 2.59 GW。目前全球投入运营的最大数据中心是中国电信在内蒙古呼和浩特的一处设施，占地约 100 万平方米。SpaceX 的 Colossus 项目占地 7 万平方米，OpenAI 支持 Stargate 项目占地 3.5 平方公里；Meta 计划在路易斯安那州建设的 Hyperion 项目占地 11 平方公里。全球数据中心的总用电量估计已超过了整个法国的用电量。