OrangeBot.AI Digest — 2026-08-19

Create design systems, build with agents

Native macOS hex editor with a brain for file formats

Looping case study videos from your images, no keyframes

The Git forge built for the age of coding agents

Code Editor, Terminal, Git, AI Agent for Android

Build Fab-ready Custom PCBs with one prompt

Summon monsters straight from your agent.

Kid friendly paper toys, free to decorate and COPPA safe.

Automate your B2B brand going viral, with new media creators

Find and remove every trace AI leaves in your text

Collaborative agents who build, learn and publish in 1 place

Let agents source clips from terabytes of your local video

ChatGPT, built differently for teens

A tiny animated cat for every browser tab.

AI agents that run in a loop

Skills have emerged as a practical and effective approach for enhancing LLM agents at inference time through structured packages of knowledge. However, existing evaluations largely measure whether skills improve aggregated task success, leaving a more fundamental question underexplored: \textbf{When do skills help, why do they work, and where do they fail?} Through controlled experiments across various benchmarks, agent harnesses and LLMs, we isolate the effects of representation, outcome annotation, retrieval difficulty, and cross-framework robustness of skills. To further answer this question, we design a contrastive study that combines controlled quantitative experiments with paired trajectory analysis. We normalize 8,135 trial records from controlled experiments and retain 238 valid unique labels from 240 open-coded records. We consolidate these observations into a taxonomy of three high-level categories and twelve skill-use modes: skills work when noisy trajectories become procedural anchors that stabilize execution. Skills improve over Workflow Memory by 6.06 points in matched comparisons. Procedural anchoring accounts for 65.7\% of skill cases, versus 4.5\% for explicit knowledge injection, showing that skills stabilize action rather than inject missing facts. Retrieval is a separate bottleneck: as pools grow from 5 to 100, actual-use precision falls from 29.6\% to 3.3\%. Confusable distractors impair offline identification, yet downstream success remains stable; exact ground-truth invocation is neither sufficient nor necessary. Skills fail under brittle assumptions, incompatible contexts, or insufficient adaptation. These findings move evaluation beyond aggregate success rates and guide reliable self-evolving agents.

Reinforcement Learning (RL) has been promising in single-turn LLM fine-tuning. However, long-horizon agentic reasoning introduces increasingly branching interactions and sparse rewards, exposing several limitations of RL: its heavyweight backpropagation-based training stack makes it impractical to fine-tune larger LLMs, and longer-horizon trajectories make credit assignment in RL substantially harder. This paper argues that evolution strategies (ES) can be a better choice for fine-tuning long-horizon LLM agents. Compared with agentic RL, ES offers three key advantages: 1) Model Scalability: ES enables full-parameter optimization with only minimal, inference-level GPU memory, making it possible to fine-tune large LLMs. 2) Flexibility: its lightweight, black-box feedback interface makes ES fine-tuning easy to compose with prompt-space evolution (e.g., skill optimization & test-time compute); and 3) Long-Horizon Scalability: ES performs trajectory-level parameter attribution without decomposing rewards across horizons, yielding better scalability than Agentic RL as the horizon length grows. Based on this insight, we propose Agentic ESOpt, a full-parameter agentic fine-tuning framework tailored to flexible parameter--context co-evolution. At each step, Agentic ESOpt samples perturbations around the current LLM parameters, evaluates the resulting agents with rewards, and applies an online reward-weighted update. To improve the exploration--adaptation trade-off, Agentic ESOpt further introduces a cosine decay schedule of the perturbation scale σ. On WebArena-Lite, full-parameter optimization of Qwen-3.5-27B improves the No Skill baseline by 6.69%. In test-time automatic heuristic design, Agentic ESOpt performs online prompt--parameter co-evolution, improving its matched baseline in 28 of 36 settings.

Artificial superintelligence (ASI) requires AI to move beyond mastering existing knowledge toward exploring the unknown, creating new knowledge, and turning new ideas into verifiable results. However, the capabilities of today's AI systems are still largely built on learning, compressing, and applying existing human knowledge. Accordingly, existing benchmarks primarily test whether AI can produce correct answers based on learned knowledge, or whether it can complete tasks under extensive human guidance. We therefore introduce ASI-Bench, the first benchmark to jointly evaluate AI systems' capabilities of innovative exploration and autonomous scientific execution across general research domains, and the first to progressively withdraw human methodological guidance within the same research project to test how far AI can proceed on its own. Built by over 40 experts with the cost of 31,000+ human hours, ASI-Bench contains 60 project-level research tasks across 11 scientific domains and progressively reduces methodological guidance to test whether AI can independently select methods, conduct research, and produce verifiable results. All tasks undergo expert review, AI-assisted auditing, sandbox execution, and scorer validation. Across 18 state-of-the-art agent--model configurations, the average score drops from 50.91 with full methodological guidance to 29.10 with only the method specified and 26.62 when agents must determine the method themselves. This sharp decline shows that current systems remain heavily dependent on human guidance and are still far from autonomously conducting end-to-end, project-level scientific research. ASI-Bench is open to the world. We invite researchers and builders everywhere to contribute new tasks, challenge the limits of today's AI, and help accelerate humanity's collective path toward artificial superintelligence at https://asibench.apexin.ai/submit.

Frontier open-weight models are increasingly available, but serving them still largely assumes datacenter infrastructure. We present FreeToken, an edge-native MoE serving system that treats a personal machine not as a small GPU, but as a unified, elastic inference platform. FreeToken co-designs the full serving stack, including model layout and loading, expert residency, CPU--GPU execution, agentic state reuse, and runtime memory management, around two realities of local AI: agent workloads continuously change their execution pattern, and edge hardware exposes heterogeneous resources whose balance differs from machine to machine. Rather than committing to a fixed offloading strategy, FreeToken continuously maps computation and model state onto the resources actually available. FreeToken supports more than 20 MoE models and real coding and tool-using agents across hardware ranging from an 8GB laptop GPU to a single workstation GPU. More importantly, it changes what these machines can practically serve, from a 35B model on a laptop to a 284B model on a gaming desktop and the 753B GLM-5.2 on a single workstation GPU. FreeToken turns open weights into deployable local software, making the machines users already own a practical platform for frontier-scale intelligence. We release the system at flashml.ai.

Although Large Vision-Language Models (VLMs) have significantly advanced embodied navigation, their direct deployment remains challenging, as existing methods often force VLMs into unnatural action spaces that misalign with their 2D pre-training priors, compounded by rigid reasoning schedules and inefficient memory management. To overcome these limitations, we propose TAMP-Nav, a unified framework for efficient embodied navigation. First, we introduce a Pixel-to-3D Action Formulation (Point) that reformulates navigation into 2D visual prompting. Specifically, the VLM merely selects 2D pixels, which are then projected into 3D coordinates for a low-level SLAM controller. This design naturally aligns embodied execution with the VLM's inherent 2D visual capabilities. Second, we propose an integrated Selective Reasoning and Anchor-Trajectory Memory mechanism (Think and Memorize), which dynamically triggers Chain-of-Thought and retains high-fidelity memory only at critical nodes, compressing redundant trajectories into lightweight Space-Time Indicators, thereby preserving critical historical information and enhancing spatio-temporal perception. Finally, we design an efficient Two-Level Alignment Paradigm (Align) via Group Relative Policy Optimization (GRPO). By superimposing global outcome rewards with fine-grained process rewards, this dense supervision tightly aligns the agent's cognitive planning with physical environmental feedback, endowing the model with adaptive reasoning capabilities. Experiments demonstrate that TAMP-Nav achieves state-of-the-art performance (e.g., 66.2% SR on R2R-CE) with high runtime and sample efficiency (requiring only 90k training trajectories).

Creative agents still lack an effective way to learn from high-quality human films, limiting their ability to produce cinematic-grade videos. A key challenge is the absence of a structured video representation that is both faithful to film content and directly usable for agentic reasoning and manipulation. To address the challenge, we propose the Agentic Video Auto-Encoder (AVA-Encoder), a framework for learning agent-native video representations via agentic auto-encoding. AVA-Encoder transforms a video into a knowledge graph (KG) representation and then reconstructs it back into video. Its hierarchy and state nodes store structured text, while a linked asset layer holds generated images, audio, and video. Typed edges preserve the relations between these text descriptions and assets in a form that agents can easily understand, query, and edit. The video reconstruction differences drive a textual-gradient optimization framework, which expresses evaluation feedback as natural-language update directions for Data-Independent Encoding Policy Pseudo-Training in the outer loop and optional Data-Dependent KG Representation Refinement in the test-time inner loop. Extensive experiments show that AVA-Encoder improves by 20.7 percentage points over the strongest external baseline. In the controlled policy-only setting, its pseudo-trained shot-level Agentic Video Encoder policy also outperforms a carefully human-tuned policy while using 74.3% fewer system-prompt tokens. We release the complete AVA-Encoder framework, a reliable agentic video reconstruction benchmark, and the first dataset of high-quality film KG representations.

High-resolution image editing is increasingly demanded in professional workflows, yet existing diffusion-based models remain constrained to resolutions below 1K due to quadratic attention complexity and prohibitive memory requirements. A prevalent workaround employs a two-stage pipeline: editing at low resolution followed by independent super-resolution. However, this approach suffers from two critical issues: information divergence, where hallucinated details contradict the original high-resolution (HR) source, and texture degradation, manifesting as over-smoothed or over-sharpened artifacts. We propose EditBridge, a diffusion bridge framework for efficient ultra high-resolution editing. Unlike conventional diffusion that regenerates from noise, we formulate refinement as structured data-to-data translation from the low-resolution (LR) edited result to its HR counterpart, explicitly conditioned on the original HR source to preserve authentic details. To efficiently incorporate HR source guidance, we introduce a prior-guided block-wise sparse attention mechanism that exploits semantic correspondence from first-stage editing to constrain cross-image interactions to spatially aligned regions, significantly reducing computational overhead. Extensive experiments demonstrate that EditBridge achieves high-fidelity editing with superior perceptual quality at resolutions up to 4K, delivering 3.6--8.4times speedup at 2K and enabling practical 4K editing in 61 seconds.

Modern agents operate inside agent harnesses that manage tools, context, and control flow, making the harness a critical part of the agent system. Our original Agent Lightning introduced a disaggregated architecture that connects arbitrary agents to RL training through an LLM endpoint proxy, an approach later adopted by frameworks such as verl Uni-Agent, AReaL 2.0, slime, and Polar. We refer to this paradigm as harnessed agentic RL, where the deploy-time harness directly participates in model post-training. Harnessed agentic RL differs fundamentally from traditional agentic RL: the harness, rather than the training engine, owns the environment interaction loop, while the trainer observes only sequences of LLM request-response pairs. This introduces challenges in retokenization, sample merging, advantage calculation, loss normalization, and backend scheduling, which can substantially affect training stability and effectiveness. We present Agent Lightning v1.0, a lightweight framework for harnessed agentic RL implemented in approximately 3,500 lines of code. It supports arbitrary agent harnesses and serves as a practical testbed for studying these challenges. We evaluate it on instruction-following, search, and coding agents, and provide a complete reproducible pipeline for coding-agent RL. Using only 6K training examples and modest compute, RL improves Qwen3.5-9B on SWE-bench Verified from 41.8% to 56.4%, a 14.6-point absolute gain. We release the complete workflow and training scripts to facilitate reproducible research on harnessed agentic RL.

Latent video generation relies on autoencoders to define a compact space in which generative models operate. Although video autoencoder architectures have evolved substantially, their latent spaces are still optimized primarily for pixel-level reconstruction and provide limited high-level semantic organization. A reconstruction-optimal latent space, however, need not be well suited to generative modeling. We propose V-RAE, a video representation autoencoder that builds compact generative latents on top of frozen vision foundation model representations. A lightweight temporal pooling module removes temporal redundancy while preserving semantic structure, and a video decoder reconstructs continuous motion from the compressed features. We evaluate V-RAE with four representative frozen encoders on video reconstruction, semantic probing, and class-conditional generation. V-RAE achieves 2.13 rFVD on K600, outperforming all evaluated large-scale pretrained video VAEs. Its latents retain substantially more semantic information than conventional video tokenizer latents. Under matched generation settings, our best variant achieves gFVD scores of 117.86 and 19.16 on UCF101 and K600, respectively, while converging up to 6x faster}. We further show that reconstruction quality alone is insufficient to characterize generative utility and introduce tFVD, a temporal-coherence diagnostic that correlates more reliably with downstream generation quality. Beyond video generation, V-RAE also improves future video prediction on Cityscapes over the Wan 2.2 VAE latent space under matched prediction settings. Taken together, the experiments show that frozen semantic representations can support video reconstruction, generation, and predictive modeling. The project page: https://v-rae.github.io/.

As text-to-image generative models advance, they raise critical safety concerns, particularly the generation of Not-Safe-For-Work (NSFW) content such as violence and nudity, further exacerbated by red-teaming adversarial attacks. Existing defenses predominantly operate under white-box assumptions, relying on text encoder optimization, weight editing, or inference-time intervention, and fundamentally cannot scale to proprietary models. Black-box alternatives based on LLM prompt rewriting offer broader applicability, yet fail in a critical regime we identify as the benign adversarial problem: prompts that are linguistically safe but still trigger harmful generation due to the model's learned data distribution. We propose DiSCO, a zero-shot, strictly black-box defense that operates entirely at the prompt level as a plug-and-play module, requiring no model retraining, fine-tuning, or access to model internals. DiSCO performs distribution-guided suffix expansion via beam search, optimized through contrastive scoring over safe and unsafe image pools generated by the target model itself, with iterative adaptive feedback until safe content is produced. We demonstrate that DiSCO consistently enhances the safety of both undefended and defended models on the I2P benchmark under multiple red-teaming attacks, achieving 37.7% and 25.13% ASR reduction, respectively, while maintaining semantic fidelity and improving image coherence. As a black-box, architecture-agnostic module, DiSCO can be readily applied to any text-to-image system without necessitating any changes to the model itself.

The quality and diversity of instruction-based video editing datasets are steadily improving, yet existing datasets mainly focus on single editing operations and fall short in supporting compositional instruction-guided video editing. In particular, multiple editing intents must be jointly understood and faithfully executed within the same video. To address this issue, we introduce CoinVE-200K, a large-scale, high-quality dataset for Compositional Instruction-Guided Video Editing. CoinVE-200K contains 1080p video-editing pairs of up to 201 frames, covering diverse compositional scenarios where each sample involves 2 to 5 atomic editing operations. The instructions target humans, objects, and backgrounds, and cover edit types such as addition, removal, modification, and stylization. All samples are built through a carefully designed generation and filtering pipeline to ensure instruction faithfulness, visual quality, temporal consistency, and compositional diversity. We also introduce CoinVE-Bench, a benchmark for compositional-instruction video editing across diverse subjects, operation types, and instruction complexities. Furthermore, we present CoinVE-Edit, a 22B compositional video editing model built upon Wan2.1-T2V-14B and Qwen3-VL-8B-Instruct. CoinVE-Edit disentangles region-aware attention for different editing instructions, enabling precise multi-region editing while preserving irrelevant content and temporal coherence. Experiments on CoinVE-Bench show that CoinVE-Edit achieves strong performance in instruction following, compositional editing accuracy, visual quality, and temporal consistency.

Byte-level hierarchical language models (LMs) have recently emerged as a robust alternative to their popular counterparts that use subword tokenization. However, generating one byte at a time remains a bottleneck for inference speed. To address this, we introduce multi-byte prediction (MBP), which generates multiple bytes in parallel, speeding up inference with minimal performance impact and no additional parameters. MBP builds on the popular multi-token prediction (MTP) paradigm with two crucial innovations. First, we introduce a variable-length prediction window that aligns with the latent tokens, or segments, of a hierarchical LM. Second, we implement a novel attention-masking scheme that enables parallel byte prediction without violating causality. We show that multi-byte prediction strikes a Pareto-optimal trade-off across multiple generative tasks, instruction following, question answering, summarization, and machine translation, achieving the best trade-off between performance and inference throughput.

Memory is becoming core infrastructure for long-horizon LLM agents, yet existing evaluations offer limited guidance on which memory substrate, namely the underlying medium in which memory is represented and stored, should be used under different operating regimes. We present a controlled harness evaluation of memory substrates for memory-augmented agents, covering dense and sparse indices, text records, structural stores, hierarchical stores, refinement-based memories, parametric updates, and activation-compatible context mechanisms. Across three backbone models and four benchmark suites spanning user-centric question answering and agent-centric decision-making, we instrument 26 performance and efficiency metrics under a unified harness. Our results show that no single substrate consistently dominates: broad retrieval benefits long-context factual QA, while excessive retrieval can harm sequential decision-making by shifting attention away from action-critical context. Scalability introduces a further routing axis, as substrates that perform well at moderate history lengths can become costly or brittle at longer horizons. These findings motivate substrate routing as a necessary component of adaptive agent memory systems and provide empirical guidance for designing efficient, reliable, and regime-aware long-term memory for LLM agents. Code will be made available upon acceptance.

Vision encoders are a critical component of vision-language models, and scaling their capacity effectively improves performance. However, dense scaling increases compute cost and inference latency. Mixture-of-Experts (MoE) architectures offer a compelling alternative, having enabled efficient scaling in LLMs, yet the MoE design space for CLIP-style vision encoders remains underexplored at State-of-the-Art (SOTA) levels. In this work, we systematically study MoE designs for vision encoder scaling and find that fine-grained MoE topologies yield substantial gains over both dense and standard MoE counterparts. We further propose an auxiliary-loss-free balancing variant for better expert utilization, and design a specialized MoE kernel to mitigate inference latency overhead. To enhance video capabilities while preserving image knowledge, we introduce frame-level distillation paired with a novel freezing mechanism. We pretrain a series of Mixture-of-Experts Vision Encoders (MoE-ViE) across a range of sizes, all consistently outperforming their dense counterparts. Our largest model matches the zero-shot performance of a SOTA encoder 1.7x its size at 76% of its latency. When aligned with an LLM, MoE-ViE surpasses all compared encoders on image and video benchmarks, including those with up to 5x more activated parameters. Code is available at https://github.com/facebookresearch/moe_vie.

Large-scale image generation has benefited from advances in data scale, quality, rebalancing, and recaptioning, yet conventional pipelines typically optimize task-specific datasets in isolation. A central challenge is not only how to curate each task-specific corpus, but also how to organize heterogeneous supervision according to the dependencies among generative capabilities. We present a capability-driven data infrastructure that couples capability-specific supervision construction with capability-aligned curriculum scheduling. Its three specialized yet interoperable data engines build complementary relational supervision for text-image grounding, inter-image transformation, and image-knowledge association, while caption experts align T2I and editing supervision across tasks and granularities. A multi-stage curriculum jointly evolves task composition, visual-concept distribution, data quality, and image resolution along the dependency order of capability acquisition, with capability-aware evaluation closing the loop through targeted retrieval, expert construction, and gap-aware resampling. At scale, the framework curates a 440M-image T2I corpus, 120M editing pairs, and over 27M image-entity pairs. With this infrastructure, we train multimodal diffusion models at two scales from scratch, with 3B and 6B sizes respectively. We conduct quantitative evaluation on CPI-Bench, along with qualitative evaluations across diverse text-to-image and editing scenarios. Experimental results present broad visual coverage, versatile rendering, and effective transfer across generative capabilities.

Lucas Shaw / Bloomberg : Sources: YouTube is offering top creators millions to post videos exclusively on YouTube for a period and will penalize those that post to Netflix concurrently —  YouTube will pay creators to post their videos to YouTube before any other service.  If creators post to YouTube and Netflix at the same time, they will face consequences.

Reuters : Former Meta engineering director Arturo Béjar testifies that Mark Zuckerberg prioritized growth and engagement over child safety on Facebook and Instagram —  A former engineering director at Meta Platforms (META.O) testified on Wednesday that CEO Mark Zuckerberg fostered a culture …

Amanda Caswell / Tom's Guide : Google announces new study tools, including a student hub, notebooks, and interactive 3D visualizations in Gemini, plus student offers for Google AI plans —  Head back to school equipped with AI tools for studying, researching and more  —  Once again, Google is giving eligible college students …

Axios : Letter: Stripe told investors January 1 marked the “beginning of the singularity”, a major inflection point in long-term trends, and H1 revenue rose 41% YoY —  Stripe on Wednesday told investors that January 1st marked the “beginning of the singularity,” which it refers …

Bloomberg : Sources: SpaceX approached Cognition about a potential acquisition, but Cognition didn't engage; Cognition CEO Scott Wu says the company is “not for sale” —  SpaceX Attempted to Buy AI Coding Startup Cognition  —  Video Player is loading.  —  Unmute  —  Current Time 0:00 Loaded: 26.60% Playback Rate

CNBC : Sources: OpenAI CFO Sarah Friar told employees at an all-hands that OpenAI “will be a public company in 2027”, or sooner if “our business continues to inflect” —  OpenAI CFO Sarah Friar told employees during an all-hands meeting on Wednesday that the artificial intelligence lab …

Bloomberg : Sources: AI chip startup Fractile is in talks to raise ~$600M at a $6.5B pre-money valuation, up from ~$1B in May, and has an initial ~$250M deal with Anthropic —  Fractile, a startup developing chips tailored for artificial intelligence use that has a deal to supply Anthropic PBC …

The Information : Source: Nvidia has discussed an investment in Mercor as part of a round valuing the startup at $20B; Nvidia paid Mercor tens of millions of dollars last quarter —  Nvidia has discussed an investment in Mercor, a data labeling provider that helps the chip designer develop its open-source AI models …

Erin Griffith / New York Times : Stripe agrees to acquire New York-based OpenRouter; a source says Stripe is paying $7.5B, with $1.5B going to the startup's founders and $6B to its investors —  The deal is significant for the artificial intelligence industry, uniting Stripe's payments business with OpenRouter's helping businesses direct their spending on A.I. models.

Dave Michaels / Wall Street Journal : The US FTC says businesses must disclose when they use personalized pricing and it will “deploy enforcement resources” against companies that do not disclose it —  Agency says businesses must disclose when they use personalized pricing and may face lawsuits if they don't

Bloomberg : Stripe Agrees to Buy AI Firm OpenRouter; No Terms Disclosed —  Stripe Inc. has agreed to buy OpenRouter Inc., an artificial intelligence startup that helps developers access and switch between AI models.  No terms were disclosed.  —  “Tokens are the central currency for companies building with AI …

Emma Roth / The Verge : Meta launches a Mac app for Meta AI and says Meta AI can now work directly with Instagram and Facebook accounts, Meta ad campaigns, and Google Workspace —  You can share your window with the new Meta AI app, as well as connect it to Google Workspace. … Meta is launching a new Mac app dedicated to its AI chatbot.

Ina Fried / Axios : OpenAI is testing Private Safety Processing, a new technique to identify misuse patterns while preserving zero data retention protections, with early customers —  OpenAI said Wednesday that it believes a new technique will allow it to safely serve its most advanced models to businesses without needing to retain their data.

Nicola M White / Bloomberg : The US CFTC bans former Alameda CEO Caroline Ellison and FTX co-founder Gary Wang from trading for five years, as part of a settlement over their roles at FTX —  FTX co-founder Gary Wang and former Alameda Research Chief Executive Officer Caroline Ellison avoided financial penalties under …

Loree Seitz / The Wrap : Nielsen rolls out changes to make its ratings more accurate, including using data from smartwatch-like devices worn by its panelists without requiring logins —  “We are relentless in our pursuit of delivering the most accurate measurement possible for our media and advertising clients,” CEO Karthik Rao says

制药公司 Merck 和 Moderna 宣布其研发的基于 mRNA 技术的实验性癌症疫苗在 III 期黑色素瘤临床试验中初步取得积极成果。逾 1100 名高危或晚期黑色素瘤患者参加了临床试验。患者的癌细胞已通过手术完全切除。实验中疫苗与 Merck 的免疫疗法 Keytruda 联合使用。相比单独使用 Keytruda，联合疗法显著延长了患者的无复发生存期，降低了癌细胞扩散到身体其它部位的风险。黑色素瘤仅占皮肤癌的约 1%，但绝大多数皮肤癌死亡病例都是由其引发的，多数黑色素瘤复发发生在初次治疗和切除后的两到三年内。

刚果周三公布的数据显示，该国的埃博拉疫情确诊病例突破 5000 例达到 5021 例，死亡病例 2378 例。始于今年五月的埃博拉疫情主要发生地为刚果以及邻国乌干达。这场疫情已成为有记录以来扩散速度最快的埃博拉病毒爆发，其规模有可能超过 2014-2016 年席卷西非的埃博拉疫情——该波疫情有逾 1.1 万人死亡。这起疫情的病毒是罕见的 Bundibugyo 毒株，目前没有针对该毒株的获批疫苗或疗法。疫情最严重的 Ituri 省已成为暴力事件频发的地区，医护人员成为袭击目标，严重制约了抗疫工作。当前的病毒致死率为 47.4%，但不同地区死亡率也有巨大差异，其中 North Kivu 省的致死率高达 70%。

当一只老鼠闯入上海动物园白颊长臂猿馆时，猴子没有试图吃掉它，或者驱赶它。一只雌性长臂猿迅速抓起了它，温柔地捧着它，抚摸着它。看到这一幕的牛津大学演化生物学家 Cyril Grueter 感到十分困惑。他随后与一个国际团队合作，发现了更多灵长类动物与其它物种亲密互动的例子。一只年轻雄性倭黑猩猩小心翼翼地抱着一只獴崽。一些猴子爬到猪背上；另一些猴子骑在鹿背上。一只灰叶猴抚摸着一只松鼠。一群短尾猕猴一起为一只流浪狗梳毛。研究团队共收集了 427 个案例，数据来源包括科学论文、媒体报道以及一项针对 37 位灵长类动物学家的全球调查。互动多数涉及玩耍或梳毛，发生在野外和圈养环境中。涉及的物种包括 88 种灵长类动物和 127 种其它物种。不是所有的动物互动记录是友好的，有 26 起事件被认定是虐待。研究发现，灵长类动物与其它物种的互动模式与其年龄和性别有关。幼年灵长类动物更倾向于玩耍，雌性灵长类动物则更倾向于梳毛。大多数互动都由灵长类动物主动发起。研究结果表明，饲养宠物的行为不是突然出现在人类身上的，而是源远流长。

童年时期创伤经历可能会导致成年后缺乏动力，从奖赏中获得的快乐减少——即快感缺失（anhedonia）。人类的童年创伤与大脑奖赏相关通路活动的改变有关，这些通路来自海马体区域。动物临床前研究表明，海马体的改变可能会导致快感缺失。但与童年创伤相关的海马体改变是否会使人更容易发展出快感缺失呢？研究人员让一群参与者填写了童年创伤和体验快乐能力的问卷，然后使用 fMRI 检查海马体与动机相关脑区的相互作用。结果发现，童年时期经历过创伤且海马体通路受损的人，比童年创伤但海马体通路未受损的人表现出更多的快感缺失。

中国科技大学等单位在远距离量子中继研究中取得重要突破，成功实现了两个冷原子量子存储器间跨越 420 公里光纤的量子纠缠，并在 230 公里以上距离突破无中继量子纠缠分发的理论极限。这一成果大幅提升了物质量子比特间的纠缠距离，为城际尺度量子网络构建奠定了基础。量子存储器间的远距离纠缠是构建量子互联网的技术基础，有望应用于基于量子中继的远程量子通信、分布式量子计算、分布式量子传感等方向。中国科大团队近年来在该研究方向持续取得突破。2020 年，团队成功实现经由 50 公里光纤的双节点量子存储器间纠缠；2024 年，在此基础上，于合肥市构建了国际首个城域三节点量子存储网络；2026 年，进一步实现了百公里高保真双节点纠缠和器件无关量子密码分发。

IEEE Spectrum 发表了一篇文章纪念了今年五月去世的数字信号处理先驱刘必治。刘被广泛认为是现代数字信号处理领域的奠基人之一，其研究推动了声音、图像和视频处理从模拟到数字的转变。尽管在工程界之外鲜为人知，但全世界数十亿人使用的技术都包含了他的研究成果，让手机通话、流媒体视频和互联网通信成为可能的低功耗数字信号处理器，其技术基础可追溯到他在 1970 年代和 1980 年代开展的研究。刘于 1934 年出生于上海，后随家人迁往台湾，他的父亲 Henry Liu Sr. 是一名电机工程师。他于 1954 年在国立台湾大学获得电机工程学士学位，毕业后随家人迁往美国，他与父亲一起就读于布鲁克林理工学院（现纽约大学 Tandon 工程学院），两人在 1956 年一起获得电机工程硕士学位。刘之后继续深造，四年后获得电机工程博士学位。他在 1962 年加入普林斯顿大学。他与其前研究生 Abe Peled 在 1976 年出版了教科书《Digital Signal Processing: Theory, Design, and Implementation》，该书是工程师的标准参考书，定义了数字信号处理这一专业。

诺基亚计划今年年底前裁减大部分在华员工，并分阶段关闭各地工厂，该公司将主要保留售后服务。前员工透露，裁员补偿采用 N+3 方案 —— 即按服务年限 N 计算的经济补偿，外加三个月工资。虽然受影响人数未公布，但诺基亚年报显示，截至 2025 年底，该公司在中国大陆、港湾地区共拥有约 7,200 名员工。据诺基亚中国官网的信息，诺基亚在中国设立了研发、市场、服务、全球交付，以及供货等全产业链布局。诺基亚在 1995 年创立北京研发中心。官网介绍，这是诺基亚全球重要的移动网络和诺基亚软件的研发基地。

蓝箭航天于 8 月 19 日 7 时 35 分在东风商业航天创新试验区发射了朱雀三号重复使用遥二运载火箭，将鸿擎科技的鸿鹄 03 星送入预定轨道；7 时 41 分火箭第一级按预定程序成功软着陆于甘肃省民勤县朱雀三号着陆场坪。这是朱雀三号火箭的第二次飞行，它成为继美国的 Falcon 9 和 New Glenn，以及中国的长征十号乙火箭之后第四种经过飞行验证的入轨级部分可重复使用火箭。本次任务是中国首次实现运载火箭一子级着陆支腿方式回收，也是中国首次实现入轨级运载火箭一子级陆地回收（长征十号乙火箭为海上网状回收）。朱雀三号高 66 米，直径 4.5 米，第一级使用了 9 台天鹊 12 引擎，第二级使用了 1 台天鹊 15 引擎，它的着陆方式与 Falcon 9 基本相同。

因为 AI 热，主要内存厂商都将产能集中到 AI 数据中心使用的 HBM 内存上，导致消费者使用的 DDR5 内存条供应短缺。过去 12 个月部分 DDR5 内存条价格上涨了 500%。今天一套 128GB 内存套装售价高达 3399 美元，是之前最低价的十倍多。一套 64GB（2x32GB）DDR5-5600 内存套装去年夏天售价不到 200 美元，如今超过 1100 美元。上一代 DDR4 内存条的价格也普遍上涨 120%-180%，虽然没有 DDR5 上涨的那么离谱。硬盘和固态硬盘的价格也普遍上涨了 125%。DRAM 芯片如今是按重量计算的全球价值最高的商品之一，主流 DRAM 芯片的每公斤价值超过纯金的一半。

美国通信公司 Comcast 正在数百万台兼容 Xfinity 网关上启用 Wi-Fi 运动感应功能，允许路由器通过测量网关和连接设备之间的信号中断探测运动。该功能是 Xfinity Shield 服务的一部分，不需要额外付费。Wi-Fi 感应技术已存在一段时间，但直到最近其准确性和可靠性才得以提升，从而得到广泛应用。Linksys 在 2021 年推出了类似的服务，但几年后便终止了。照明公司 Wiz 在 2023 年推出了一系列支持 Wi-Fi 感应的智能灯泡，而飞利浦 Hue 最近也在其产品中部署了类似的射频感应技术（使用 Zigbee 而非 Wi-Fi）。Comcast 用户可根据自己的活动开启或关闭该感应功能。

二次大战结束后，英国实施了数年的糖供应配额制，这一限制于 1953 年 9 月结束，居民的糖摄入量随即大幅增长，糖消耗量几乎翻了一番。这一历史提供了罕见的机会了解糖摄入量对未来健康的影响。研究人员分析了 1951-1956 年间出生于英国的逾 64,000 人数据。较早出生者在出生后前 1000 天内受制于更长的糖摄入量限制，而较晚出生者受到限制的时间较短。研究人员利用英国生物银行的数据跟踪了参与者数十年的健康信息。研究人员发现，出生前 1000 天内经历较长时间糖摄入量限制的人群罹患五种癌症的几率较低：乳腺癌、前列腺癌、肝癌、直肠癌和肺癌。其中肝癌的降低最为显著，发病率降低了约 69%；乳腺癌的降低幅度最小，但也降低了 36%。他们还发现糖分限制与生物衰老相关，研究人员通过端粒长度衡量生物衰老，在出生前 1000 天内经历更长时间糖摄入量限制的人的端粒更长，这种差异相当于生物衰老速度减缓了约 2.2 年。出生后很少摄入糖的人在成年后也会延续这一趋势，他们的总糖摄入量更少，饮食也更健康、更多样化。研究人员强调，这一结果并不意味着儿童完全不应该吃糖。

Mozilla 释出了 Firefox 154。主要新特性包括：Local Network Access 保护支持 WebSocket 连接，当网站尝试与本地网络的一个设备建立 WebSocket 连接，它们需要先获得许可；Smart Window 能建议相关标签页分组，并为每个分组建议一个名称；Windows、Linux 和 macOS 都支持本地配置文件备份，能在所有三个平台上恢复备份；网站可设置在浏览器关闭时豁免清除其 Cookie 和网站数据；地址栏新增 Manage AI 快捷操作；修复 bug，等等。

Google 以 1000 万美元拍下了破产航空公司 Spirit 的数据。Spirit 成立于 1980 年，主要经营加勒比海地区、拉丁美洲等定期国际航线。该公司在新冠疫情后遭遇财政困难，2026 年 5 月宣布倒闭，公司进入清算程序，目前正拍卖资产以筹集资金和偿还部分债务。法庭文件显示，Google 以 1000 万美元拍下 Spirit 的数据，包括 1 亿封电子邮件、5 亿条 Microsoft Teams 信息、1700 万份 OneDrive 文件，2050 万 SharePoint 文件，逾 3000 万则客户服务电话录音和逾 1500 万则客户服务聊天记录，60 万个 ServiceNow 工单，以及逾 76.3 万次航班、500 万组机组人员、逾 120 万张燃油单据，787452 个零部件采购记录等公司运营数据，等等。Google 表示购买这些数据是为了改进其 AI 服务。法庭文件显示，这些数据在出售前已进行匿名化处理，Google 也承诺会清除其中发现的任何个人身份信息。

根据在 macOS Tahoe 26.7 RC 版本发现的一则视频，苹果正在研发配备摄像头的耳机 AirPods，该产品处于即将发布状态。AirPods 的摄像头可以在 AI 帮助进行目标识别，在演示视频中，苹果通过摄像头识别一本书的书名。摄像头会将信息传输给 Visual Intelligence‌，Siri 将能回答佩戴者周围环境的问题并记录信息。配备摄像头的 AirPods 代号为 B790，最早可能会在 9 月宣布。

根据 FDA 上月发布的年度调查报告《2025 National Youth Tobacco Survey》，2025 年美国青少年吸烟率降至历史新低的 1.4%。该年度报告自 1999 年以来一直在追踪 11-18 岁美国青少年吸烟情况。美国青少年吸烟率历史最高水平发生在 1997 年，当时有 36% 的高中生表示正在吸烟，70% 表示至少尝试过一次吸烟。美国青少年的电子烟使用率也在 2025 年降至 5.2%。美国成年人的吸烟率保持相对稳定。老年人的吸烟率仍然高于年轻人，他们面临的吸烟相关风险最大。吸烟率的下降可能主要归功于公共卫生政策和关于吸烟危害的公共宣传。但最近 Make America Healthy Again 运动将尼古丁宣传为一种健康产品，未来吸烟率是否回升有待观察。