Weekly Digest — 2026-W42

Create multi-agent workflows that you can trust

Chat with visitors from your website or link

Hand-drawn diagrams your AI agent can draw with you

The AI that finds money you're missing and gets it back

Create explainers, screencasts, and AI avatar videos

Delegate coding. Get back code that works.

Autonomous assistant designed to handle meetings and email

See how simulated buyers react, right in your agent

The open-source personal agent that runs on your computer

Books and PDFs read aloud by neural voices, on device

Bring your ideas to life through the camera.

Orchestrate agent fleets with custom tools built on macOS

Online reinforcement learning (RL) fine-tunes diffusion models with black-box rewards, but reverse-process methods such as Flow-GRPO are limited in efficiency by poor credit assignment, while forward-process methods such as DiffusionNFT are efficient yet unstable without carefully tuned exponential-moving-average (EMA) off-policy training. We show that NFT-style objectives contain a hidden forward--reverse consistency force that joint positive--negative training cancels only partially, and that EMA stabilizes training through restoration rather than off-policy rollouts. We introduce DiffusionART (Diffusion Reinforcement with Aggregation-First Reward Tilting), which aggregates each prompt's rewards into a shared reward-improving direction for reverse-process states, removing the consistency force by construction, and adds a trust-region-masked second update for drift-adaptive restoration without EMA tuning. It requires no differentiable reward model, remains effective with fewer samples per prompt, and supports pairwise reward models (DiffusionART-Pairwise). Across image and video generation, DiffusionART trains efficiently and stably without per-task tuning. In particular, on HunyuanVideo-13B with equally weighted visual and motion quality rewards, DiffusionART reaches the late-stage mean reward levels of DiffusionNFT and Flow-GRPO-Fast with 12.02 and 11.72 times fewer GPU hours, respectively, while achieving 28.8% and 53.0% higher mean training rewards.

Multimodal language models (MLLMs) exhibit promising spatial reasoning capabilities, yet the internal mechanisms supporting them remain poorly understood. We uncover S-Space, a linear representation subspace in which object-token representations encode continuous coordinates along horizontal, vertical, and distance axes. Although we identify these axes by differentiating spatial-answer scores with respect to object-token activations, S-Space is intrinsic rather than induced by specific data or prompts, remaining readable under non-spatial questions. We show that S-Space functions as an internal spatial workspace of MLLMs. Perception encodes object locations into it, forming spatial representations that can be directly read out to support accurate spatial judgments. Internal reasoning transforms it for changing viewpoints. Verbal responses selectively access it, evidenced by causal interventions that alter spatial judgments while largely preserving non-spatial responses. Case studies further show that S-Space assigns locations to unseen entities and the observer, and connects physical directions with abstract linguistic meanings, suggesting both the capabilities and limitations of large-scale language training. Finally, we use S-Space as a diagnostic tool for spatial reasoning failures. Direction-word readouts reveal systematic coupling between cardinal and image-relative reference frames. More importantly, a perception-computation decomposition exposes a key bottleneck: MLLMs can form capable spatial representations, yet fall short in the spatial transformations required to reason reliably across viewpoint changes.

Humans effortlessly react to the world; they adeptly return a tennis shot, catch a falling cup, step aside from a passing cart, or pass a box tossed during truck unloading. These everyday actions require anticipating a moving target and reflexively responding with the whole body within about a second. Such visual reactive whole-body control remains challenging for humanoids, which must (1) coordinate whole-body control, (2) react to object dynamics in real time, and (3) ground perception in egocentric vision. We study catching thrown boxes as a representative task that demands all three capabilities; existing systems cover only part of it, relying on end-effectors, motion capture, or marked objects. We present Reflex, a three-stage framework that learns these capabilities stage by stage while connecting them through a compact representation of object dynamics. First, a state-based policy learns whole-body catching with reinforcement learning. Second, the policy learns to infer the dynamics of a moving box from delayed and incomplete observations. Third, an egocentric RGB-D encoder learns to recover this dynamics representation directly from visual observations while keeping the whole-body controller fixed. This decomposition provides each capability with a direct training signal and makes visual learning substantially more scalable. In simulation, Reflex achieves 85.7% catch success from egocentric RGB-D, compared with 87.9% when given privileged box observations. Temporal visual history is critical: removing it reduces success by 35.6 and 39.8 points on centered and laterally offset throws. On a real Unitree G1, Reflex catches human-thrown boxes with 65% and 40% success at 2.5 and 5 m, reaching 76% and 59% of human catchers' success at the same distances.

Computer-use agents (CUAs) are rapidly improving, yet synthesizing diverse and reliable software environments at scale remains challenging. Native application environments require substantial construction and integration effort, which limits application coverage and increases the risk of application-specific overfitting. Browser-hosted environments provide limited interaction depth and workflow complexity. These limitations motivate SoftwareGym, an automated pipeline for synthesizing GUI software environments with diverse interfaces, executable functionality, and runtime-verified behavior. It generates structurally diverse designs, realizes them as runnable applications, and uses GUI evaluation with iterative repair to ensure functional reliability. Using SoftwareGym, we synthesize 182 applications across twelve categories and achieve a mean GUI task pass rate of 91.5%. We construct SoftwareGym-Bench, a 120-task evaluation suite spanning the same twelve categories, and use trajectories collected from these environments for CUA training. The resulting model improves Pass@3 from 41.39% to 56.97% on the strictly out-of-domain OSWorld benchmark and achieves 40.0% on SoftwareGym-Bench. These results demonstrate the potential of synthesized software environments as scalable and transferable supervision for CUA training.

Multimodal large language models have become the dominant paradigm for visual understanding, but incur substantial costs by encoding inputs into dense, fixed-size patch tokens. However, visual information is unevenly distributed: some regions require fine-grained detail, while others admit compact representations. Downsampling sacrifices this detail, while existing token pruning and adaptive approaches remain limited in content-adaptive granularity, task generalization, and integration with modern MLLMs and serving infrastructure. Overcoming these limitations calls for foundation models that learn, end to end, where-and at what granularity-to allocate visual representations, a native capability we term elastic visual representation weaving. We introduce VisionWeave, establishing this capability in frontier-level MLLMs through large-scale training. It combines two components: a gated spatial pooler constructs coarse-grained representations alongside native fine-grained representations within a shared MRoPE coordinate, while a granularity router learns their content-adaptive allocation. Through self-distillation alone, we validate this capability on Qwen3.5-4B and scale to Qwen3.8-27B with over 30K A100 GPU-hours. Based on Qwen3.8-27B, VisionWeave adaptively adjusts token savings to visual content, saving 43.0% tokens on average while retaining 98.9% native performance across eight benchmarks, versus only 88% performance preserved for token pruning baselines with a fixed 50% savings target. Extensive evaluations confirm robust efficiency-quality trade-offs across diverse tasks, resolutions and video frames. When deployed on SGLang serving engine, our method achieves a 2.3x throughput gain while reducing mean TTFT by 54.4% and mean TPOT by 60.6%. Together, we believe these results position elastic visual weaving as a promising capability for next-generation multimodal models.

Adversarial games have driven advances from heuristic search to reinforcement learning, yet learning and adapting strategies from limited samples remain challenging. AI agents offer an alternative by turning game experience into revisions of executable policies. Building on heuristic learning (HL), we formalize Adversarial Heuristic Learning (AHL), a paradigm that uses AI agents as learning engines to refine game policies and supporting software while keeping model weights fixed. We introduce AAArena, a benchmark comprising 12 authentic adversarial games and 1,920 archived human programs, with an evaluation protocol modeled on real-world game competitions. Agents interpret rules, choose opponents, analyze replays, and revise game agents to achieve their highest ranking within fixed match and evaluation budgets. We evaluate completedmodels model and harness configurations: Opus5.5 with Claude Code earns 6 gold medals, while no evaluated configuration tops the remaining 6 human ladders. Performance is generally weaker in games with more complex rule specifications. Further experiments show that opponent selection and dense feedback support policy improvement, and that agents learn from both on-policy replays of their own matches and off-policy replays of other players' matches. These results highlight HL's potential in adversarial games and identify persistent challenges in game understanding, strategy implementation, and long-horizon policy development.

Precise grounding matters. It specifies which object is the target and where that object is, even in clutter and for tiny objects, and it has to be fast enough for closed-loop control. Yet vision-language-action (VLA) and world-action models (WAMs) take perception from general-purpose vision-language and video-generation backbones, which still fail in these settings. We introduce GroundingPI, a 4B grounding foundation model that generates points and boxes as quantized coordinates in a shared vocabulary. Training combines multimodal and spatial pretraining, supervised fine-tuning, and reinforcement learning with GRPO, using supervision from public datasets and dedicated data engines. Against 44 baselines across 34 grounding benchmarks spanning 11 perceptual capabilities, GroundingPI establishes a new state of the art, averaging 73.68%, above the larger GPT-6 Astra (71.54%). As a downstream visual backbone, GroundingPI improves performance on robotic manipulation and autonomous driving. On RoboTwin 2.0, it outperforms every mainstream backbone we evaluate in all four out-of-distribution settings, by up to 24.8% relative to the strongest backbone. On RoboCasa-GR1, GroundingPI trained with 50% of the demonstrations outperforms those baselines trained with 75%. On nuScenes, used as the visual backbone, GroundingPI attains an average open-loop L2 error of 0.296 m. We systematically analyze GroundingPI's pretraining in scale and data composition. Downstream autonomous driving and robotic manipulation improve as the pretraining is scaled. Analyzing the data recipe across these 11 perceptual capabilities shows dense grounding's substantial benefits for both, and OCR's potential as a catalyst for perceptual learning. These results support grounding as a perceptual foundation, and dedicated perceptual pretraining as a promising direction for foundation models of physical intelligence.

Large Language Models (LLMs) increasingly ship with explicit "thinking modes", yet their counterpart, "no-thinking", has received far less attention. We study LLMs' no-thinking behavior along two axes. a. How to measure no-thinking? Prior work typically defines no-thinking through proxies such as a disabled thinking mode or the absence of long traces. These proxies are unreliable: disabled thinking modes may still emit reasoning, while long traces may contain filler rather than genuine inference. We instead normalize each response into a pre-answer trace and final answer, and evaluate it at three levels: (i) Empty-Thinking Rate for strict answer-only compliance; (ii) instruction-aware Question-Pre-answer Relevance for similarity between the question and pre-answer trace; and (iii) LLM-as-judge Explicit Inference Rate for visible explicit inference. Together, these metrics distinguish answer-only output, relevant but non-inferential text, and explicit inference. b. How does no-thinking vary across tasks and models? We evaluate six prompting interventions on six LLMs across Boolean, multiple-choice, and open-ended questions. We find that explicit no-think controls cannot reliably eliminate visible inference. Models instead exhibit "Thinking Inertia": explicit inference persists even under strict controls and becomes more prevalent as the answer space opens. Accuracy remains stable on Boolean and multiple-choice tasks, whereas open-ended tasks reveal a trade-off between answer-only compliance and task accuracy. Rewriting the same questions across answer spaces shows that supplying candidate answers makes answer-only responses easier to produce. These findings establish no-thinking as a non-trivial capability: stopping explicit reasoning cannot be assumed from model settings or instructions alone and deserves systematic evaluation alongside reasoning ability.

As the production of software passes into the hands of software itself, coding agents are increasingly commissioned to build agents fitted to their users' own workflows. This delegation carries a quiet assumption, that the delivered agent will faithfully serve the requirement it was built for, rather than merely appear to. No existing benchmark tests that assumption. We introduce AgentSWE, a benchmark for agent software engineering spanning three stages of the agent development lifecycle: Creation of a complete agent from a natural-language requirement, Editing of production agent codebases, and Optimization toward a measurable target. Each of its 25 tasks is framed as a commission rather than an exam: the builder iterates against a handful of illustrative cases, while acceptance exercises the frozen submission on held-out cases designed to span the full requirement, grading behavior against requirement-derived rubrics under hard pass/fail gates or, for Optimization, against the source benchmark's own objective. Evaluating a range of coding agents across models and harnesses, we find that none comes close to delivering what a requirement asks for: the gap between what a builder demonstrates and what it delivers is the norm, not the corner case. An accompanying empirical analysis of the failure trajectories identifies recurring failure modes, among them overfitting to the visible cases, hollow runtimes behind code that looks complete, and unsupported claims of completion, each a signal by which coding agents can learn to build the agents they are asked for.

Terminal agents are moving beyond local operations toward completing real-world professional tasks. Short trajectories can demonstrate commands and tool calls, but they do not fully show how agents accumulate evidence through exploration, use it to inform judgments, continually adjust strategies based on feedback, and turn intermediate results into final deliverables. High-quality long-horizon trajectories capture these agentic behavioral patterns and their connections within a single problem-solving process, providing richer experience for sustained autonomous work. However, environment synthesis may complete or simplify the key work underlying these patterns in advance and thus remove the need for exploration, implementation, and verification that the agent was meant to perform. To address this, we introduce TerminalHorizon, a fully automated data synthesis engine for long-horizon terminal agents. Starting from real-world professional work, it uses independent execution to identify key work omitted during construction and uses the resulting evidence to guide environment reconstruction. To extend these patterns across stages, it progressively expands tasks under the same goal, introducing new requirements that build on prior results and call for further exploration and improvement. Using this engine, we construct 1.5K task environments and perform supervised fine-tuning of Qwen3.5-27B on the resulting dataset to obtain TerminalHorizon-27B. We achieve substantial performance gains on all 12 benchmarks spanning coding-agent tasks, agentic tasks in real-world production settings, and general reasoning. Multiple RSI case studies, each with a 12-hour research budget, further demonstrate the model's ability to continue experimenting beyond an initial solution, improve its methods based on feedback, and retain effective improvements in its final deliverables.

Video interactive storytelling enables viewers to actively steer how a video unfolds. However, once we allow viewers to intervene during generation, a new challenge arises: The viewer's request can have latent dependencies on both the grounding source and the current rendered video state. These dependencies may not be explicitly stated in any individual input, but emerge only when the source, rendered history, and new viewer intent are considered jointly. Existing interactive video generation systems primarily emphasize following viewer instructions, while source-grounded video generation methods focus on aligning generated content with an external narrative or knowledge source. This leaves a fundamental question underexplored: What context should a generation model ground on during interactive continuation, and how can heterogeneous, unstructured inputs be transformed into such grounding context? In this work, we formulate contextual grounding as the process of transforming heterogeneous input context into an executable constraint model for video generation. To address this challenge, we introduce VIS-Ground, which performs Structured Context Abstraction to recover grounded states and cross-context dependencies, Generation Constraints Induction to project relevant dependencies into candidate-specific constraints, and Constrained Video Generation to enforce these constraints through planning, verification, revision, and rendering. Across three video generation backbones, VIS-Ground consistently achieves the highest overall composite score, reaching an average absolute improvement of 10.3 points over the strongest per-backbone baselines. Detailed analysis further shows gains across both narrative and knowledge grounding, and reveals remaining challenges in dependency extraction, and faithful realization during video rendering.

Autoregressive (AR) grounding models serialize spatial predictions, introducing sequential latency and imposing a causal order on output tokens. We view grounding as visual evidence extraction: objects, locations, and spatial relations are jointly constrained by the image and query, yet their dependencies do not imply an intrinsic left-to-right generation order. This distinction makes bidirectional diffusion a natural fit, allowing spatial hypotheses to emerge in parallel and be jointly refined through iterative denoising. We introduce GroundAnything, a 4B-parameter grounding foundation model that reconciles fast parallel decoding with precise localization through blockwise denoising. Training combines grounding pretraining from public datasets and dedicated data engines, direct AR-to-diffusion conversion with joint AR and diffusion objectives, supervised fine-tuning, and GRPO-based reinforcement post-training. Across 30 grounding benchmarks, our autoregressive variant, GroundAnything-VLM, establishes a new overall state of the art among similarly sized models at 72.42%, remaining competitive with GPT-6 Astra (71.35%). With entropy-guided decoding, GroundAnything also surpasses the prior state of the art at this scale, averaging 61.75% versus 53.32% for the fast MTP-based LocateAnything model. We further explore decoding strategies, showing that an optional self-speculative mode achieves a 4.51times speedup over the AR counterpart with a 0.74 percentage-point drop in COCO F1mIoU. Infrastructure experiments show that progressive inference optimizations translate parallel decoding into practical speedups. These support efficient visual grounding in latency-sensitive real-world systems.

Claire Ruckin / Bloomberg : Sources: JPMorgan and Goldman, leading a $5B loan sale with a yield of ~11% for AI startup Volta, took a rare move to limit exposure by underwriting just $1.3B —  JPMorgan Chase & Co. and Goldman Sachs Group Inc. have taken a rare move to limit their exposure on a $5 billion loan sale …

Anthropic : Anthropic researchers say small AI models may be months away from being able to self-replicate on IoT devices like smart ovens and off-brand Android TV boxes —  Our experiments suggest that small AI models may be months away from being capable enough to self-replicate on commercial IoT devices …

Jason Schreier / Bloomberg : Sources: Epic has delayed the release of an upcoming game, instructing developers to implement an “AI first” production model with “no more hand-written code” —  The company told the team behind the upcoming shooter that it will now be an ‘AI first’ production

Brock E.W. Turner / Axios : Suki, an AI ambient scribe for healthcare providers, raised a $90M Series E led by Healthier Capital; a source says it projects $60M-$80M in revenue this year —  Suki, a health care AI and scribe company, raised a $90 million Series E round to scale and add agentic workflows, CEO Punit Soni tells Axios Pro exclusively.

Cara Lombardo / Wall Street Journal : Sources: Bumble is in advanced talks to be taken private at a ~$1B valuation by Ithaca Holdings, founded recently by two Silicon Valley execs; BMBL jumps 49.61% —  Founder Whitney Wolfe Herd expected to stay involved with company as part of deal  —  Bumble , the dating app founded by Whitney Wolfe Herd …

@artificialanlys : Muse Wonder Full (max) outperforms Fable 5.1 (max) on Artificial Analysis' Intelligence Index, matches Sonnet 5.5 (max), and is just shy of Opus 5.5 (max) —  Meta has released Muse Wonder and Muse Wonder Full, achieving Artificial Analysis Intelligence Index v4.3 scores of 52 and 56 at max reasoning effort - each a material gain over Muse Spark 1.3 (max) at 48, with a release gap of only 6 weeks Muse Wonder is available for users on Meta's first-party ...

The Information : A look at OpenAI's safety concerns over loop transformers before shipping its next major model; Microsoft accidentally confirmed GPT-6.1 Sol uses the technique —  OpenAI is grappling with fallout from its use of a controversial innovation in AI design as it works to get its next unreleased model …

Annelise Levy / Bloomberg Law : Apple v. OpenAI: a US judge partly grants Apple's motion for expedited discovery, ordering OpenAI to hand over forensic images of devices and other documents —  Apple Inc. convinced a federal judge in California to force OpenAI Inc. to expedite turning over documents concerning its alleged theft …

Dustin Volz / New York Times : The FBI says it worked with CrowdStrike to dismantle 40+ laptop farms across Texas used to help North Koreans get remote US tech jobs under stolen identities —  F.B.I. investigators worked with the cybersecurity firm CrowdStrike to dismantle so-called laptop farms that steal American identities to earn cash for Pyongyang.

Jon Hicks / GamesIndustry.biz : Xbox CEO Asha Sharma announces Xbox Labs, a small team to test emerging ideas, citing recent game ports and mashups, many of them built with AI tools —  New business unit is charged with testing new projects and judging them by user feedback, spurred by recent mashups and ports delivered with AI coding tools

The San Francisco Standard : Federal prosecutors say a Saudi national was arrested over bugging an OpenAI executive conference room; OpenAI found the audio recorder around September 30 —  An OpenAI engineer has been arrested on suspicion of hiding an audio recording device in an executive conference room …

@openainewsroom : OpenAI says the FBI arrested one of its staffers after the company found a hidden recording device in its offices and used its models to understand the device —  We discovered a sophisticated recording device that was hidden inside one of our offices. We used our models to understand the device and where it had come from, identifying that a specific employee had planted it a few days earlier. We referred the matter to the FBI, and we're grateful to the agen...

加州公司 Atomic Canyon 与核工业组织合作推出了针对核反应堆运营的 AI 助手 NIVA——代表 Nuclear Industry Virtual Assistant。该 AI 助手已进行试点，目前向美国全部 94 座核反应堆开放使用。另一家核能领域的 AI 初创公司 Nuclearn 表示其产品已服务于逾 65 家美国合作伙伴，其中包括传统核电站，也包括开发先进小型核反应堆的公司如 NuScale Power。Nuclearn 首席财务官兼联合创始人 Jerrold Vincent 指出，AI 是解决核工业最紧迫难题的理想工具。核反应堆运营面临的难题包括了繁琐的监管规定以及复杂的管理与维护系统，这恰恰是 AI 擅长处理的任务。核反应堆的运营需要处理大量文书工作。核电站会产生海量文档，用于证明符合监管要求，记录针对各种潜在问题的应对过程。Nuclearn 的 AI 产品能帮助核电站更高效检索数据库，更快向联邦政府提交文件，或查找过去是如何解决特定问题的，它称得上是核能领域的 ChatGPT。

亚马逊最近更新了一系列自有品牌产品线，其中自有品牌平板产品发生了重大变化。过去 15 年亚马逊一直销售 Fire 平板电脑，如今 Fire 品牌名被放弃，亚马逊改为推出 Alexa 平板，而且运行的是经过 Google 认证的 Android 操作系统。以前 Fire 平板运行的 Fire OS 是亚马逊定制的 Android 版本，亚马逊还提供了自己的应用商店，如今应用商店已经关闭，Alexa 平板运行的是完全 Google 化的 Android 操作系统，以及 Google Play 商店和其它常规应用。Alexa 平板共三款：Alexa Tablet 8 起售价为 230 美元，Tablet 11 起售价 330 美元，Tablet 12 Pro 起售价为 500 美元。前两款使用了联发科入门级处理器 8189，第三款使用了联发科的天玑 8400。该平板的一大亚马逊特色是用户可通过唤醒词 Alexa 与 AI 机器人进行对话。这些平板将于 10 月 14 日上市。

2026 年诺贝尔经济学奖授予了斯坦福大学的 Robert E. Hall 和哥伦比亚大学的 Michael Woodford，以表彰两人突破传统凯恩斯主义框架，从家庭和企业等微观主体行为出发建立经济模型，证明无数家庭和企业的日常决策最终会塑造整个经济运行方向。这套理论如今已成为全球央行决策的重要基础，也成为应对衰退、通胀和能源危机的重要工具。数十年来对凯恩斯模型的应用已暴露出其局限性，包括未能有效抑制失业，反而推高通货膨胀。

一项对数百家公司编程实践的研究发现，人工代码审查是 AI 编程工具整体效率的一大瓶颈，导致几乎没有证据表明企业使用 AI 工具能提高软件产量或减少就业。研究涵盖了 2021-2026 年 3 月间逾 700 家软件开发公司逾 70 万员工的 3 亿条工作事件（如 commits 和 pull requests）以及问题管理软件数据。公司在引入 AI 编程智能体后，代码总行数平均增加了 30%，commit 总数增加了 20%，pull requests 总数平均增加了 23%。但代码行数的增加并没有直接转化为软件产出的提升。原因在于代码审查流程，引入 AI 编程智能体后，代码审查的平均耗时明显延长。总体而言，引入 AI 智能体后，代码审查流程时间平均增加了 49%。引入 AI 智能体后，代码审查的员工人数增加了 14%。虽然 AI 智能体贡献了部分代码审查意见和 pull requests，绝大部分审查工作仍然是由人类完成的。

早在 1970 年代，面对石化产品在全球暖化中的作用，美国石化行业制定了一系列策略，旨在推迟气候行动以维护自身经济利益。行业组织 American Petroleum Institute 在 1980 年发布了一份政策手册，歪曲了科学研究结果并淡化全球变暖的严重性。研究发现，2008-2018 年间美国石化行业组织在与气候变化相关的政治活动上总共投入了约 34 亿美元。牲畜业排放的甲烷对全球暖化也有重要影响。发表在 PLOS Climate 期刊上的一项研究发现，美国主要畜牧业行业组织 National Cattlemen’s Association(NCA)采用了与石化行业相似的策略。美国在很长时间里是全球最大的牛肉生产国，最近几年被巴西超越。该组织早在 1989 年就知道了牲畜对全球变暖的贡献，但一直致力于淡化甲烷的排放问题、质疑气候科学，以及反对减少牛肉消费。

东京的一名 40 多岁的男子习惯于睡前启动 BitTorrent 下载成人视频，第二天早上关闭软件。但从今年五月起他先后收到了两封信函，勒令他支付 50 万日元否则将面临民事诉讼。他一开始以为是诈骗，但很快发现是真的。越来越多的日本人因为使用 BitTorrent 下载成人视频而收到罚单。东京地方法院知识产权部门的数据显示，它收到的披露身份信息的申请数量从 2023 年的 814 起增至 2024 年的 2454 起，2025 年进一步达到 5320 起。2024 年主要 ISP 机构收到的信息披露请求中 95.6% 都涉及成人视频的版权侵权指控。相关诉讼显示，成人视频制作公司引入了能识别涉嫌通过文件共享软件传播受版权保护内容的用户的 IP 地址的系统。由于相关纠纷增加，日本通信部门已建议用户不要随意使用 BitTorrent 之类的软件。

众所周知，工蜂蜇人之后会将螫针留在皮肤上自己飞走，而螫针会自行穿透皮肤组织泵送毒液。发表在 PNAS 期刊上的一项研究利用高速摄像技术揭示了蜜蜂螫针如何泵送毒液的过程。工蜂留下螫针的现象被称为自切（autotomy），自然界里相当罕见。螫针由三个互锁部件组成，共同构成了排毒管：一个向外扩展形成主泵室的管心针（stylet），以及两根尖锐的、可前后滑动的刺针（lancets），它们刺入组织并将毒液注入目标。研究人员利用镊子取下活工蜂的螫针，将其插入人工凝胶块，利用体视显微镜和高速彩色摄像技术捕捉了排毒动作。视频显示，连接在每根刺针上的瓣膜在前后滑动泵送时会改变形状。它们在向下冲程时像降落伞一样充气，通过类似活塞的动作形成密封，对泵室施加向下压力以挤出毒液；在向上冲程时则会瘪下去。当这些瓣膜在中间相遇时，它们的形状会再次改变，以便能彼此错开通过。

《巴黎协定》对全球气温上升幅度设定了一个比工业化前水平高 1.5°C 的阈值。现有气候模型预测这一阈值会在 2030 年代突破。但根据《npj Climate and Atmospheric Science》期刊上发表的一项研究，突破时间可能比预期的更快。对截至 2025 年的气温记录的分析显示，2011-2025 年间的全球变暖速率达到了每 10 年上升 0.40°C，比 1971-2025 年间每 10 年暖化速率 0.20°C 翻了一倍。这意味着 1.5°C 的阈值会在 2028 年突破，比此前的预测早了十年。暖化加速主要归因于东亚-北太平洋、欧亚中纬度地区以及南大西洋-南极洲区域变暖趋势的加剧。

代表美国游戏业的行业组织 Entertainment Software Association (ESA) 向美国贸易代表办公室提议将最近炙手可热的游戏破解黑客 voices38 列入恶名盗版市场名单。匿名黑客 voices38 在短时间内发布了一系列使用 Denuvo DRM 的游戏破解补丁，其中包括《霍格沃茨之遗（Hogwarts Legacy）》和《黑神话：悟空》。ESA 的理由是，游戏破解黑客在盗版生态系统中扮演着重要角色，没有破解，盗版网站就无法提供可运行的盗版游戏。ESA 认为将其列入名单有助于版权所有者打击破解活动。

著名基地建造与管理游戏《RimWorld》开发商 Ludeon Studios 宣布释出 v1.7 版本，将游戏引擎从 Unity 切换到自研的 Waylith Engine，性能提升了 2-4 倍。Ludeon Studios 称，游戏的外观和玩法完全一致。相比 v1.6 版本，v1.7 在启动速度、加载大量模组时的速度、地图生成、世界生成、每秒更新次数（TPS）以及几乎所有其它方面都有显著提升。开发商表示，一个多达 1500 个模组的模组包加载速度从 14 分钟缩短到 6 分钟。虽然引擎变了，但存档都能向前向后兼容，v1.6 的存档能在 v1.7 中正常加载，反之亦然。模组开发者无需为 v1.7 修改模组，只需要测试之后标记下兼容 v1.7 即可。

美国曾在 2000 年宣布消灭了麻疹，但因为虚假信息和接种率下降，麻疹疫情正卷土重来，今年至今麻疹病例总数已达到 4080 例，创 1991 年以来的最高纪录。其中宾夕法尼亚州的疫情最为严重，该州今年已经累计报告 1136 例病例，六人死亡。死者包括一名新生儿、一名 6 周大的婴儿、一名 18 岁青年、一名 20 岁青年、一名 40 岁成年人以及一名 39 岁女性。所有死者都未接种疫苗，两名婴儿尚未达到接种疫苗的年龄。宾夕法尼亚州的疫情无放缓迹象，过去一周新增 157 例病例。美国现任卫生部长 Robert F. Kennedy Jr. 是坚定的反疫苗者。

美国太阳能板生产商表示，他们即将迎来一项技术突破，这项突破可能颠覆整个行业，甚至威胁到中国在该行业的垄断地位。这些制造商正在竞相开发新型太阳能板，其发电量比目前市面上大多数同类产品高出 25%。如果成功，这一进步将使世界能以更少的土地和屋顶面积来满足迅速增长的电力需求。新技术被称为叠层太阳能，通过像层层蛋糕一样添加第二层和第三层材料，使太阳能电池板能够将更多的太阳辐射转化为电能。研究人员多年来一直致力于完善叠层太阳能电池板，包括 Tandem PV、First Solar 和韩华 Qcells 在内的几家公司表示，在成功测试这些设备并获得技术机构认证后，他们正准备很快将此类电池板投入商业销售。但新的叠层太阳能板及其制造商将面临许多挑战。电力生产商将不愿购买过于昂贵的新设备，除非制造商能让他们相信，叠层太阳能电池的性能衰减速度不会比传统太阳能电池更快。