OrangeBot.AI Digest — 2026-08-31

Sell ad space on your laptop

Create and edit precision motion graphics & video with chat

Drive the full SDLC with AI agents, step by step

See what new sites are built with, the week they launch

Turn public buying signals into sales pipeline

A ball for boring meetings to keep you busy

The programmable mic you can squeeze, shake & play

Learn the terminal like a game — in a browser sandbox

Record your iOS & Android screen from your Mac.

Practice speaking with AI.

The Podcast Search Engine

The complete digital wardrobe experience

Create cinematic screen recordings on your Mac

Planner that turns goals into daily progress

Create custom stamp shapes

Loop Engineering is emerging as a practice for organizing development work around coding agents. Instead of writing each prompt by hand, practitioners design loops that monitor progress, assign work, run checks, and decide what the agent should do next. Even with a capable coding agent, a loop may trust a stale progress note, skip needed verification, spend its budget in the wrong direction, or stop before the task is safe to submit. Yet the final outcome of one end-to-end run cannot tell whether success or failure reflects the loop's guidance or the coding agent's ability to carry out the task. We introduce LoopArena, a benchmark for evaluating how well one model can guide a separate coding agent through a long-running task. The model under evaluation is the Controller: after each coding round, it receives a structured summary of the run and instructs a separate, fixed coding agent, the Worker, on what to do or verify next, or decides whether to stop. LoopArena evaluates this ability in three complementary settings that differ in execution scope and cost. Type I scores next-step Loop Contract selection through execution-validated questions without running the Worker at evaluation time. Type II executes repeated control over a selected slice of a full task, while Type III evaluates the paired full task from its original state. On full tasks, the best observed Strict Success Rate is 24.69\%, leaving substantial room for improvement in long-horizon loop control. Across Controllers, the paired reduction in estimated inference cost averages 64.4\%, and Type II produces a similar ordering under the main Core criterion (Spearman's \(ρ=0.9747\)). We release the benchmark data and evaluation code at https://github.com/AMAP-ML/LoopArena .

Equipping Large Language Models (LLMs) with multi-turn tool-calling capabilities is essential for building autonomous agents. However, progress is fundamentally limited by the reliance on full-length trajectory imitation. For tasks involving multiple order-independent sub-goals, the optimal solution space forms a vast combinatorial diamond lattice. Forcing this rich topology into monolithic trajectories causes a severe topological collapse, indiscriminately penalizing valid alternative explorations and severely degrading policy diversity. To address this, we propose DART-SD (Diamond-topology Aware Retrieval and Tuning for Self-Distillation), a novel framework that shifts the paradigm from global forcing to topology-guided localized correction. DART-SD first models the execution process as a converging Interaction-State Transition Graph (ISTG), faithfully capturing the inherent diamond topology of successful and failed exploratory paths. During autonomous rollouts, the framework identifies the Critical Topological Breakpoint (CTB) and retrieves success-supported recovery references. Finally, we introduce a progressive self-distillation paradigm through CTB-guided localized supervision, ensuring that the training loss is calculated exclusively on the generated recovery steps while strictly protecting the valid reasoning prefix from destructive gradient updates. Experiments on complex multi-turn tool-calling benchmarks demonstrate that DART-SD significantly outperforms traditional full-trajectory baselines.

Scaling robot data is crucial for building generalist Vision-Language-Action (VLA) models, yet robot trajectories are harder to scale than web-scale image-text data because embodied collection is costly and sparsely covers the physical world. This makes representation quality a central bottleneck: under a fixed robot-data budget, continued pre-training must turn limited trajectories into transferable visual-action knowledge rather than merely fit actions. We propose VLAct, a VLA-oriented VLM backbone trained on broad, heterogeneous, multi-embodiment robot data before task-specific fine-tuning. VLAct preserves the broad VLM prior and encourages shared action semantics across embodiments through VLM-prior preservation, multi-head continuous action co-supervision, and a partially unified cross-embodiment action layout, while allowing task-specific action heads during fine-tuning. Across simulation, real-world, and unseen-embodiment transfer, VLAct consistently improves downstream performance under fixed fine-tuning protocols. On LIBERO-Plus and RoboTwin 2.0, VLAct surpasses industrial VLA systems including ABot-M0 and LingBot-VLA, achieving success rates of 82.6% and 92.5%. On RoboDojo, VLAct ranks sixth among all policies by success rate and outperforms all explicitly designated world-action model (WAM) entries on both metrics. Most notably, on RoboCasa-GR1, an unseen humanoid embodiment, VLAct using only 20% of downstream trajectories outperforms the full-data GR00T-N1.6 baseline. These results are obtained using fully open-source data and only a 16-GPU training setup, showing that representation-centric continued pre-training can deliver highly competitive performance under a modest compute budget and is an important independent axis of VLA progress beyond data scaling.

Generative models can turn natural-language prompts into images, text, code, and other content, lowering the cost of producing drafts and components. Their practical impact increasingly depends on whether those pieces can become complete, dependable deliverables. This survey examines agentic artifact creation, which we define as stateful construction in which an AI system materially constructs or revises a deliverable and intermediate observations redirect later work. Functionally, the process links an operational representation of the artifact, a construction policy, and runtime verification whose feedback can redirect later actions. We reviewed 259 works available through August 20, 2026: 230 systems meeting this definition and 29 benchmarks of agentic artifact construction. We compare six artifact families, then analyze application settings and evaluation practice as separate dimensions. Across families, construction challenges reflect not only modality but also how tightly decisions are coupled and whether failures become visible while they remain repairable. Decomposition can reduce local complexity while increasing coordination and reassembly costs. Learned judges may add little independent evidence when they share the generator's preferences or blind spots. We formulate principles for keeping commitments and responsibility explicit, turning feedback into targeted repair, and revalidating affected state after change. We also identify opportunities for sustaining coherent, accountable control as artifacts, creator intent, and construction systems evolve. A curated paper list is available at https://github.com/GeminiLight/awesome-agentic-artifact-creation.

Physical understanding and reasoning depend on forming compact and generalizable representations of the world. While modern vision-language models can recognize and explain diverse physical events, they often lack explicit representations of the underlying mechanisms-such as object states, physical parameters, and governing dynamics-needed for reliably reasoning how the world evolves and responds to interventions. In this work, we introduce Code-as-World, a paradigm that represents physical worlds through executable world representations. By expressing physical composition, dynamic evolution, and visual appearance as executable code, Code-as-World provides a compact, quantitatively grounded, and controllable abstraction of the physical world. To construct such representations from multimodal observations, such as natural-language descriptions or real-world videos, we develop an agentic discovery loop inspired by abductive reasoning, where an agent proposes, executes, renders, verifies, and iteratively refines executable world hypotheses. As a concrete application, we use verified executable worlds to provide scalable physical supervision for training vision-language models on quantitative physical reasoning. Experiments show that Code-as-World-VL achieves state-of-the-art performance on QuantiPhy and surpasses leading proprietary models, highlighting the potential of executable world representations as a scalable foundation for physical intelligence.

Self-evolving language models have recently emerged as a promising path toward superintelligence, with the advantage of reducing the cost of human supervision. While considerable progress has been made in verifiable domains, self-evolution in unverifiable domains remains substantially less explored. We propose Judge co-adaptation from Zero data (J-Zero), a unified Challenger--Solver--Judge co-evolution framework that supports self-improvement across both domains. The Challenger and Solver co-evolve through an adversarial interaction: the Challenger generates increasingly difficult tasks, while the Solver learns to produce higher-quality responses to them. In parallel, the Judge co-adapts using preference pairs whose ordering is known in advance from how each response was produced, i.e., the Solver's answer over the Challenger's, and its decomposed-and-recombined answer over its one-shot answer, rather than from the Judge's own scores. J-Zero outperforms the baselines by an average of 4.2 points on verifiable and 8.0 points on unverifiable domains, and continues to improve through at least ten iterations, whereas the baselines degrade after two.

Streaming 3D reconstruction from extremely long videos requires estimating camera motion and scene geometry online under bounded memory and computation. Early streaming models achieve causal, bounded-cost inference using finite context buffers or compact recurrent states, yet their estimates often deteriorate as sequences grow. Recent methods improve long-horizon stability by coupling short-range context with persistent or multi-level long-range memory. We pursue a different route: we keep the learned temporal state strictly local and formulate predictions whose targets remain independent of sequence length. We present ABot-Recon, a simple streaming model that caches KV features from only the preceding 11 frames. It predicts a point map in the current camera coordinate system together with an adjacent-frame relative pose. These predictions remain equivariant under changes of reference frame, and global poses and geometry are recovered through sequential composition. To reduce accumulated drift, a lightweight temporal refiner improves relative rotations using recent visual and motion context, while a composition-aware pose loss supervises multi-step pose composition. Extensive evaluations on challenging long-sequence benchmarks demonstrate the superior long-horizon performance of our local-context approach. On Oxford Spires, ABot-Recon achieves an ATE of 4.35 m and an RPE-R of 0.12^circ, reducing both errors by approximately 40\% relative to the best prior results.

Language model pretraining has become almost synonymous with prohibitive cost, placing it out of reach for much of the academic and open-source communities. Although strong open-source efforts already exist, including open-weight models and open-source training recipes, a cost-efficient, hardware-accessible, and open-source pretraining recipe has long been missing. Even at a small scale, training Llama-3.2-3B costs over \1.5M, and reproducing SmolLM3-3B needs over 700K. In this report, we present an open pretraining recipe designed to lower this barrier. Using this recipe, we train a collection of Puro-2B models from scratch on up to 1.4 trillion tokens with FP8 precision on consumer-grade RTX 5090 GPUs. The models in the collection differ in token budgets and selected recipe variants. Our best model is trained at a compute cost of less than \6.9K and approaches Qwen2.5-1.5B performance under our evaluation protocol. This cost efficiency is enabled by a combination of approaches, including hardware selection, low-precision training, hyperball optimization, curriculum model averaging, and the data recipe. Beyond the recipe itself, we provide two additional results. First, across the Puro-2B collection, we derive a Puro Cost Scaling Law that relates training cost to average model performance; the fitted law suggests that about 4.4K, less than \$5,090, is sufficient to reach the performance of Qwen2-1.5B. Second, as an end-to-end case study, we examine how pretraining data curricula shape downstream performance after post-training. Such controlled studies are enabled by having access to the full pretraining pipeline rather than model weights alone. We release the full training recipe for Puro-2B, including data, code, and model weights under Apache 2.0 at https://huggingface.co/collections/thu-pacman/puro-2b.

Vision-Language-Action (VLA) models can turn multimodal context into robot actions, but their action decoders are still trained largely by behavior cloning. This supervises which motor command was demonstrated while leaving implicit the local objective served by the behavior under the instruction. Future-based supervision enriches action learning with frames, latent observations, trajectories, or motion representations, but these signals capture particular realizations of what may happen rather than the shared semantic objective of the forthcoming behavior. We propose Intention Distillation (INDI), which distills behavior-level intent into the action decoder. During training, a frozen teacher VLM interprets a demonstrated segment from the current observation, instruction, coarse action summary, and corresponding execution video. From its standard inputs, the deployed VLA recovers the resulting multimodal intent representation at an intermediate decoder layer and uses it to organize action prediction together with representations of how the behavior unfolds and what it achieves. On SimplerEnv-Bridge, INDI improves GR00T-N1.7 from 64.3% to 84.7%, and on RoboCasa Kitchen it improves the controlled GR00T-N1.7 baseline from 64.1% to 70.3%, with consistent gains on π_{0.5} across both benchmarks. In real-world tasks, INDI improves average success from 62.0% to 68.7%, with gains of up to 12.0 pp on longer-horizon tasks. Further analyses show that the recovered latent is used by the decoder, captures behavior objective and execution progress, and organizes downstream predictions in an objective-dependent manner. These results show that action decoders benefit from explicitly modeling the semantic objective of the behavior they generate.

Autoregressive video diffusion enables scalable long-video generation by producing chunks from a bounded recent context. While recency-based caching preserves local continuity, it evicts historical cues needed when subjects, objects, scenes, or attributes reappear. Existing memory mechanisms expose models to nonlocal history, but access alone does not ensure effective use. Our analysis reveals that video DiT layers exhibit distinct preferences for current, recent, and distant context, suggesting that long-range memory requires deciding both what to retrieve and where to use it. We introduce LayerRecall, a current-conditioned, layer-selective memory router that retrieves relevant historical K/V states and injects them only into backbone-specific memory-sensitive layers while preserving local attention elsewhere. To reduce reliance on scarce high-quality long-horizon videos and explicit memory-allocation labels, we further propose Cross-Horizon Prediction Matching (CHPM), which uses a privileged long-context reference to supervise the bounded-memory router in prediction space. Across 100 multi-shot evaluation prompts, LayerRecall achieves the best overall results on MemoBench and MovieBench while matching its backbone on VBench-Long, demonstrating stronger long-range recovery without sacrificing local continuity. Qualitative analyses further reveal memory-guided self-correction, whereby initially mismatched local attributes return to their historical appearance without resetting ongoing motion or scene structure. Additional analyses show cross-backbone portability and negligible inference overhead.

Long-horizon agentic tasks require large language models (LLMs) to iteratively retrieve, integrate, and maintain dispersed information across multi-turn interactions, but preserving all interaction histories leads to a continuously growing working context. Recent proactive context management methods allow models to edit their own working context with specialized tools, yet they still face three key limitations: (1) a limited toolset restricted to search, deletion, and summarization, with no support for global planning, long-term memory, and adaptive compression; (2) inefficient exploration that treats context management actions uniformly despite their heterogeneous impacts on final outcomes; and (3) coarse-grained credit assignment that assigns the final trajectory-level reward to all intermediate context editing actions during RL. To bridge these gaps, we introduce ContextPilot, a proactive context management framework for long-horizon agentic reasoning. Our approach systematically augments the toolset with planning, long-term memory, and soft context offloading tools. We further propose an RL method tailored for context management, which uses context and entropy variation to identify critical editing decisions for branch sampling and estimates action-level advantages from all branched trajectories that pass through the corresponding context editing action. Experiments on long-context QA and deep search tasks show that ContextPilot achieves stronger performance with a more compact working context, consistently outperforming existing baselines across various base models and benchmarks. Code is available at https://github.com/Tencent/ContextPilot.

Spatio-temporal video grounding (STVG) requires models to identify when a referred event occurs and localize the target entity throughout that interval. Existing multimodal large language models typically serialize dense localization trajectories autoregressively, causing decoding latency to grow with tube length and allowing localization errors to propagate across time. We introduce Parallel Tube Decoding (PTD), a generative formulation that decomposes grounding into a temporal block followed by time-conditioned spatial blocks decoded simultaneously. This removes both token-level and trajectory-level dependencies, reducing the sequential decoding depth to a fixed 1 + 1 rounds, independent of tube length. To enable parallel spatial generation, we introduce Decoupled Block Attention, which preserves access to shared video-query context while eliminating cross-box dependencies, together with localization-aware policy optimization for temporal boundaries and spatial geometry. On VidSTG, PTD reduces Tube Completion Latency by 79x and increases spatial decoding throughput by 92x over standard autoregressive decoding, while also improving grounding accuracy. With a compact 4B backbone, our model performs favorably well on VidSTG and HC-STVG, and generalizes zero-shot to temporal grounding, grounded VideoQA, and referring video object tracking. Our results show parallel tube generation is an efficient and effective alternative to autoregressive localization in videos.

Recurrent fast-weight memories and selective state-space models compress an expanding context into a fixed-size recurrent state, making the state transition an online learning rule. We study this rule under read-after-write autoregressive semantics. For the prefix-prediction objective considered here, the local fast-memory example revealed at step t is the prefix-aligned pair (x_t,y_t)=(ϕ(k_{t-1}),v_t). The common same-step association (ϕ(k_t),v_t) remains causal, but optimizes a different internal objective. We derive normalized first-order updates for squared-error regression and negative inner-product objectives. The regression family comprises Falcon-1 (a scalar NLMS update), Falcon-2 (its per-column extension), and Falcon-3 (a sliding-window mini-batch update); Falcon-1A/Falcon-2A/Falcon-3A are the corresponding inner-product variants. We provide recurrent, masked-parallel, and chunk-parallel forms, together with numerically stable positive-decay renormalization. Representative variants remain competitive in language modeling and improve length extrapolation on variable-digit addition. This framework separates temporal alignment, plasticity, forgetting, and bounded rehearsal in recurrent sequence models.

Factual question answering (QA) typically assumes a single canonical answer, obscuring whether large language models (LLMs) retain divergent accounts of long-tail facts. To address this gap, we introduce ElephantBench, a closed-book knowledge probe comprising 1,094 questions generated through an auditable graph-based pipeline. The pipeline retrieves related documents from a low-exposure web corpus, identifies naturally occurring disagreements, and converts them into multi-account QA records. Each answer is verified against the originating documents and authoritative public web sources and is then reviewed by human annotators. Across 32 models, even the strongest model recovers both accounts on only 52.4% of questions, while on nearly all remaining questions it recalls one account but omits the other. Scaling model size and inference-time reasoning improve recall but do not eliminate this incompleteness. Corpus analysis further shows that exposure imbalance favors the dominant account, whereas greater minority-side exposure is associated with more complete recall. These findings establish ElephantBench as a reproducible knowledge probe for diagnosing epistemic myopia in parametric memory. More broadly, our graph-based benchmark construction pipeline provides an efficient and scalable way to turn long-tail corpora into source-traceable knowledge probes, supporting efforts to evaluate and advance the epistemic rigour of next-generation LLMs. Code is available at https://github.com/Tencent/ElephantBench.

Evaluation is shifting from static QA toward agentic settings where models act through external tools. We identify a critical yet underexplored capability within this space - dexterous visual tool use: fine-grained, closed-loop parameterized visual action in which models infer tool parameters from visual evidence, and those parameters directly govern the final result. Existing benchmarks cover web navigation, GUI operation, and software engineering, but rarely target this coupling between visual evidence and execution precision. We propose EASEL, a benchmark evaluating a controlled instance of dexterous visual tool use that adopts reference-guided visual reconstruction as its primary proxy task: the agent incrementally paints a canvas to match a reference image. EASEL additionally includes semantic tasks spanning region annotation, handwriting, and path planning. We further provide EASEL-Data, a 440k-sample two-stage curriculum dataset for trajectory supervision, and EASEL-9B to investigate its effect on this capability. Evaluation of 25 models reveals that current multimodal agents systematically struggle on EASEL. Reconstruction similarity bottlenecks at low levels (0.40-0.54), while trajectory diagnostics expose severe closed-loop instability - models typically saturate early or degrade post-peak. Semantic tasks reveal sharp capability boundaries in precision annotation and path planning. EASEL-9B, trained on EASEL-Data, surpasses the base model by a relative 6.3%, ranking third among all evaluated models.

Ryan Gillespie / Orlando Sentinel : Florida says it is revoking permits for Flock cameras and similar license-plate readers on state roads and no longer intends to issue such permits —  Florida is revoking permits for Flock cameras and similar license-plate readers on state roads, dealing a blow to local law enforcement agencies …

Chance Miller / 9to5Mac : Filing: Apple claims a former iPhone engineer used a confidential Apple circuit schematic in his work at OpenAI, evidence is being destroyed, and more —  Apple has filed a new document in its ongoing lawsuit against OpenAI as it continues to push for expedited discovery.

Lucinda Shen / Axios : Sources: AI sales and marketing startup Clay is raising a round led by Wellington at a $7B pre-money valuation, up from $5B via an employee tender in January —  Clay, an AI startup for sales and marketing teams, is raising new money led by Wellington at a $7 billion pre-money valuation, Axios has learned.

Kirsten Korosec / TechCrunch : The Pentagon launches ChatGPT Mil and Grok for Government on its GenAI.mil platform, giving its 3M personnel access to AI tools “tailored to warfighter needs” —  The Pentagon has launched versions of OpenAI's ChatGPT and xAI's Grok, giving 3 million civilian and military personnel access …

About Amazon : Amazon says the FTC's complaint over Sponsored Ads cites no evidence of consumer price increases or advertiser harm, and misunderstands how advertisers operate —  The FTC today filed a misguided lawsuit claiming Amazon misled advertisers about its Sponsored Ads pricing and auction.

Amanda L. Gordon / Bloomberg : Sources: Polymarket will raise $1B led by 1789 Capital at a $21B post-money valuation; Donald Trump Jr. is a partner at 1789 and adviser to Polymarket —  Venture capital firm 1789 Capital is leading a new funding round for Polymarket that will give the prediction-markets platform a $21 billion post-money valuation.

Annie Palmer / CNBC : The FTC and 22 state AGs sue Amazon, alleging it overcharged advertisers over $20B since 2019 via hidden surcharges, with costs largely passed on to consumers —  The Federal Trade Commission on Monday sued Amazon, alleging the e-commerce giant “secretly and systematically overcharged” …

Chance Miller / 9to5Mac : On his last day as CEO, Tim Cook pens an emotional memo reflecting on his time at Apple and says he takes “enormous comfort” in handing the helm to John Ternus —  Today is Tim Cook's last day as Apple CEO, with John Ternus set to take over tomorrow.

Dave Michaels / Wall Street Journal : FTC to File Lawsuit Alleging Amazon Deceived Advertisers —  Commission and more than 20 states to allege e-commerce giant reaped billions by manipulating ad prices on its retail platform  —  WASHINGTON—The Federal Trade Commission plans to file a lawsuit against Amazon.com on Monday alleging …

Kris Holt / Engadget : Sony launches Live TV on PS5 in the US, letting users stream 100+ free, ad-supported TV channels with content from Amazon MGM Studios, Crunchyroll, and more —  Ad-supported streaming channels are hitting Sony's console in the US today.  —  PlayStation 5 owners in the US will be able to stream free …

Cheyanne M. Daniels / Politico : Trump says communities that reject data centers will end up “backwards and poor” and that “China could not be happier with this anti Data Center movement” —  President Donald Trump on Monday again championed the spread of data centers despite growing bipartisan backlash ahead of the midterms in November.

Rocket Drew / The Information : Reframe, which uses AI and industrial robot arms to make modular homes in a factory, raised a $40M Series A extension round led by Energy Impact Partners —  For all the talk of robots that look like humans, the next robots to move into people's houses could look more like ducks.

Bloomberg : Sources: Hyperliquid is in talks to enter the US market through Kraken parent Payward, weeks after Trump said he was working to bring the platform onshore —  Hyperliquid Labs is in advanced talks to bring its perpetual futures to US traders through Kraken's parent company Payward …

Chance Miller / 9to5Mac : Gurman: Apple is moving App Store to services unit run by Eddy Cue, who previously oversaw it until 2015; Cue is taking on a more influential role under Ternus —  Phil Schiller is stepping down from his role leading the App Store and Apple events, according to a new Bloomberg report.

Fox News : The Trump administration launches Project Watershed 250, a six-month cybersecurity pilot to protect Texas water systems from cyberattacks —  FIRST ON FOX: The Trump administration is launching “Project Watershed 250” Monday, Fox News Digital exclusively learned, deploying American artificial intelligence …

自由软件视频编辑器项目 OpenShot 释出了 v4.0 版本。主要新特性包括：新色彩视图；新录制视图：将麦克风、屏幕、Web 摄像头和系统音频直接添加到项目中，每个音源保持独立且可编辑；10 种新特效；使用本地大模型选择和跟踪对象；更简洁的原生时间线；更快的特效和编辑速度；智能的创意工作流程；扩展 Qt 6 支持，改进了与较新 Linux 发行版的兼容性，为 Android 和其它平台奠定了基础。

加州参议院和众议院批准了年龄验证法案 Assembly Bill 1856。在递交给州长批准之后法案预计于 2027 年 1 月 1 日生效。法案豁免了 Linux 和 BSD 等开源操作系统。法案要求，如果操作系统有账户设置功能，那么系统提供商须提供一个界面，在账户设置期间要求输入设备主用户的出生日期、年龄或两者兼有。操作系统通过相对一致的实时 API 向受监管的应用商店和应用开发商提供数字年龄信号。该信号不显示精确的出生日期，而是四个年龄段之一：13 岁以下、13-15 岁、16-17 岁或 18 岁及以上。对 2027 年 1 月 1 日之前的设备，操作系统提供商必须在 2027 年 7 月 1 日之前提供界面让账户持有人提供所需的年龄信息。

Linus Torvalds 宣布释出 Linux 7.3-rc1，关闭了 7.3 的合并窗口，正式版预计将在十月底释出。Linux 7.3 的主要特性包括：Ryzen AI Halo LED/RGB 驱动、继续即将推出的 AMD Zen 6 的支持工作、KSMBD 兼容 Apple Time Machine 备份、内核驱动初步支持 2026 年款 Steam Controller、合并 FailFS、Intel Xe3P Nova Lake 集显支持稳定、改进了显存容量有限的系统的游戏性能、改进 SMP 降低延迟提升实时性能、等等。

上周末 Steam 平台逾 12TB 数据泄露，涵盖了该平台 2003-2013 年之间几乎所有的游戏。这些数据是通过一个公开访问的 API 获取的，但不清楚是近期访问还是早就下载但直到上周才公开。相关数据来自被称为 Steam2 的内容分发系统，2013 年 Steam2 被 SteamPipe 系统所取代，因此数据仅限于 2013 年前。泄露的数据包括了Valve 和第三方发行商发布的热门游戏的早期版本、原型版本和试玩版本，其中包括《传送门2》的被删减内容，被取消的《半条命2：第三章》的部分文件。

Google 过去一年对其搜索结果的展示方式进行了两次重大改变。其一是搜索结果链接，以前你将鼠标悬停在搜索结果上会在浏览器底部看到网站链接，现在显示的是 google.com/goto + 一串看起来随机的字符串。搜索结果中的 AI Overview 引用的链接也是采用此类展示方式。其二是用户以前可以在搜索词末尾添加 &num=100，可以在一个页面上显示前 100 个搜索结果，如今这一快捷方式被取消了，Google 强制只展示最多 10 个搜索结果，意味着你想要看前 100 个结果需要点击 10 次。

科学家早就知道，植物叶片表面分布着许多微小的气孔。当温度升高时，这些微小的孔隙会“张嘴”，让水分蒸发，从而带走热量，就像人出汗能降温一样，但气孔这一植物“散热器”背后的分子调控机制，一直是个未解之谜。该通路的核心是一种名为“泛素特异性蛋白酶24”（UBP24）的蛋白质。他们发现，当高温来袭，植物体内的激酶会“唤醒”UBP24，导致其分子电荷发生变化，使其变得更加稳定。在更稳定的状态下，UBP24 有助于“保护”并激活其他参与维持气孔开口的蛋白质，使植物的蒸发冷却系统在热应激时保持活跃。通俗来说，UBP24 就像“空调”上灵敏的温控开关。类似的故事并非只在植物身上上演。科学家还发现，啤酒酵母中的一种相关蛋白质也依靠相似原理应对热应激。酵母与植物分属不同物种，生活方式大相径庭，且两者之间存在数亿年的进化差距，却在细胞层面使用了相似的“散热逻辑”。

研究人员借助基因组证据，还原了数千年来人类食用昆虫的模式。研究表明，在欧洲、中亚与东亚地区，吃昆虫可能只是偶然行为；而在热带地区以及尼安德特人中，食虫则更为普遍。研究人员检测了 745 份来自现代人的牙结石样本，其年代最早可追溯至 3.3 万年前。牙结石能保存食物的 DNA，为研究人员提供了远古饮食的记录。研究结果显示，生活在欧亚大陆北部的现代人并不会经常食用昆虫。研究团队还检测了与分解几丁质相关的基因，几丁质是昆虫外骨骼的主要成分。在欧亚大陆北部人群中，几丁质酶基因发生了突变，导致人体消化昆虫外骨骼的能力下降。这种基因模式已延续了约 9000 年，可追溯至农业兴起之时。尼安德特人的情况则截然不同，他们的牙结石中所含的昆虫 DNA 要多得多。尼安德特人牙结石中最常见的基因痕迹来自双翅目昆虫，包含苍蝇和蚊子，其中蚊子的 DNA 含量尤其丰富。该结果佐证了近期的一个假说——尼安德特人可能经常食用带有蝇蛆的动物尸体。而蚊子遗骸的大量存在也支持了另一种观点——猎物的尸体有时可能被存放在池塘或沼泽环境中，而蚊子会在这些地方产卵。

被称为 Cloud 9 的星系可能是人类发现的第一个没有恒星的“失败”星系。此类星系虽然有大量气体和尘埃，却几乎没有恒星存在。Cloud 9 距离地球 1400 万光年，靠近旋涡星系 M94，它包含一团巨大的氢气云，估计质量是太阳的 100 万倍，以及质量约为太阳 500 万倍的暗物质。它几乎不发出任何星光。无恒星星系形成的主流解释与紫外背景辐射相关，在宇宙早期的再电离时代后，紫外辐射场将低质量暗物质晕中的气体加热到足够高的温度，使得气体无法有效冷却并坍缩形成恒星。

韩国准备向所有民众提供免费 AI 服务 AI for All。该服务计划从下月起进行 beta 测试，计划今年晚些时候推出。服务由韩国两家最大的电信公司以及 Kakao 牵头的三个联盟提供，政府提供部分算力，包括提供最多 512 块英伟达 B200 芯片和支付部分运营费用。AI 服务预计将与政府系统连接，而不只是作为独立的聊天机器人运行。居民将能使用这些服务预约医生、搜索公寓和获取税务指导。小型企业将能计算税款和查询是否符合政府补贴资格。家长将收到教育内容的推荐。每个联盟都计划推出自己的 AI 应用，在现有产品中加入生成式 AI 功能。在政府支持的计划下，用户可以无限次访问，没有 token 限制。

柏林市长 Kai Wegner 表示，本月初黑客入侵了柏林市政府系统，窃取了数据，并提出了勒索要求。他表示政府不会屈服于勒索要求。他没有透露赎金金额。据媒体报道，黑客索要 30 个比特币，价值约 200 万欧元。攻击导致部分在线系统被迫关闭。在俄罗斯和东欧活动的 Rhysida 组织宣布对此负责。该组织宣布计划一周后拍卖从柏林市窃取的 5.79TB 数据。柏林市政府官员表示，数据泄露发生在 8 月 7-12 日之间。8 月 14 日，两个政府部门的网络被关闭，导致住房补贴申请和付款在数日内无法进行。

在墨西哥湾之后，Google 宣布其地图服务 Google Maps 将安大略湖（Lake Ontario）更名为美国湖（Lake America）——这一更改主要针对美国用户，加拿大用户仍然看到的是安大略湖，而世界其他地方的用户则会看到两个名字。另一家美国地图服务 MapQuest 表示不会改名，但会提供一款工具，允许用户根据自己的意愿随便改名字。

NASA 的 Nancy Grace Roman 太空望远镜于 8 月 30 日 7:26 使用 SpaceX 重型火箭 Falcon Heavy 从佛罗里达肯尼迪太空中心 39A 发射台发射升空。太空望远镜以 NASA 首任天文学部门女主任的名字命名，使用了美国国家侦察局捐赠的 2.4 米口径主镜，配备了两台科学仪器——3 亿像素多波段红外相机大视场仪表（WFI）以及能直接观测邻近恒星周围的类木行星的日冕仪（CGI）。其核心任务包括探测暗能量、发现系外行星及验证广义相对论宇宙时空曲率。火箭在飞行 31 分钟后与望远镜分离，助推器安全返回发射场。Roman 望远镜将飞往日地之间的 L2 拉格朗日点，它将首先进行为期三个月的调试，NASA 计划 2027 年初发布望远镜拍摄的首批图像。望远镜每天将传回 1.4TB 的数据。

全球唱片巨头索尼和华纳对 Anthropic 提起诉讼，指控其犯下了历史上规模最大、最明目张胆的知识产权盗窃罪行之一。诉讼指控 Anthropic 非法利用数万首版权音乐作品训练其模型。Anthropic 及其创始人 Dario Amodei 和 Benjamin Mann 被控肆无忌惮的大规模非法下载、抓取和传播受版权保护的作品，目的是开发、运营该公司的 Claude 系列 AI 模型，并从中牟取暴利。唱片公司要求为每部侵权作品索赔最高 15 万美元，每次可识别版权信息被删除的情况则追加最高 2.5 万美元赔偿。如果法院裁决唱片公司胜诉并判决最高赔偿金额，总赔偿金额可能高达数十亿美元。诉讼还指控 Mann 使用 BitTorrent下载了逾 500 万本盗版图书，Anthropic 员工还从 Pirate Library Mirror 网站下载了逾 200 万本盗版图书，从付费获得唱片公司授权的 MusixMatch 和 LyricFind 等网站抓取歌词。

Android 安全加固项目 GrapheneOS 发现，Google 新一代旗舰智能手机 Pixel 11 取消了对硬件 MTE（hardware memory tagging）的支持，导致该项目无法完成对 Pixel 11 的支持。MTE（Memory Tagging Extension）是 ARMv8.5-A 架构引入的安全特性，通过标记分配的内存去跟踪非法内存操作，改进内存安全性。Google 是从 2023 年发布的 Pixel 8 起开始支持硬件 MTE。但 Android 和 Pixel OS 从未默认启用 MTE，相比下苹果的 iPhone 17 默认启用了它的 MTE 实现 Memory Integrity Enforcement(MIE)。GrapheneOS 会自动为更多应用启用 MTE，为每个安装的应用提供一个开关供用户可选启用。对于不兼容的应用则提供开关可选禁用。GrapheneOS 正与摩托罗拉合作推出支持 GrapheneOS 的手机，新手机将使用高通的骁龙 8 Elite Gen 5，该 SoC 支持硬件 MTE。GrapheneOS 项目不推荐用户购买 Pixel 11，建议购买更便宜的 Pixel 8、9 和 10。

在 OpenAI 之后，另一家美国 AI 关联公司 X/SpaceX 称，有约 200 个中国关联水军账号在社媒上悄悄煽动美国民众的反数据中心情绪。相关账号的推文内容包括 AI 如何加剧电网压力并推高电价，以及“描绘数据中心运营商如何以牺牲公众利益为代价中饱私囊的漫画”。前 Twitter 通信主管 Jim Prosser 反驳了硅谷关于反对数据中心是中国心理战的说法，他认为大型科技公司利用中国心理战的说法忽视当地民众的合理担忧。 “如果 Greg Abbott 和 Kathy Hochul 都能在某件事上达成一致，那这很可能不是中国的心理战。”共和党籍的德州州长 Greg Abbott 以及民主党籍的纽约州长  Kathy Hochul 最近都限制了新数据中心在当地的开发。Prosser 称，科技行业需要花更多时间倾听受影响社区的声音，而不是居高临下对他们置之不理：“如果你是生活在 Atherton 或 Menlo Park 的风投家，从未去过俄亥俄州，却对俄亥俄州居民的感受指手画脚，那就有问题了。”