OrangeBot.AI Digest — 2026-05-26

Fast, accurate STT for production-grade voice agents

AI coding platform built specifically for the Python stack

Tell Scribe what to say. It writes the rest.

API for agentic multi-document processing

Launch and run any compute job on AWS with 1 command

Turn talking, pics, or scribbles into clean, structured data

Meeting reminders that actually make you smile.

Talk, spec, ship: get your product ideas into production

Like Claude design for email marketing

A spatial alternative to Cmd+Tab for macOS

Manage Your DNS from the Command Line with DNSimple CLI

A new SOTA for compact open models on the edge

Prototype hardware visually, export real C++ for engineering

The first electric Ferrari designed by LoveFrom

Generating uncanny AI avatars is now open source

Reinforcement Learning has become a standard paradigm for aligning Large Language Models with human intent and task requirements. While Group Relative Policy Optimization offers an efficient, value-model-free alternative to Proximal Policy Optimization, adapting it to real-world multi-reward settings remains challenging. Standard scalarization practices, such as Reward Combination and Advantage Combination, suffer from significant drawbacks: Reward Combination frequently generates advantages with excessively large squared magnitudes that lead to training instability, while Advantage Combination relies on static hyperparameters and ignores cross-objective correlations. To address these limitations, we propose Dynamic Variance-adaptive Advantage Optimization (DVAO), which dynamically adjusts combination weights based on the empirical reward variance of each objective within a rollout group, effectively up-weighting objectives with a stronger learning signal while suppressing noisy ones. We mathematically prove that DVAO maintains bounded advantage magnitudes for stable training and introduces a self-adaptive cross-objective regularization mechanism. Extensive experiments on mathematical reasoning and tool-use benchmarks using Qwen3 and Qwen2.5 models demonstrate that DVAO significantly outperforms baseline methods, achieving a superior multi-objective Pareto frontier and robust training stability.

Interactive world models are advancing rapidly, yet existing benchmarks cover only part of the required competencies, leaving no unified standard for systematic evaluation. To fill this gap, we introduce WBench, a comprehensive multi-turn benchmark for interactive world model evaluation along five dimensions, namely video quality, setting adherence, interaction adherence, consistency, and physics compliance. WBench contains 289 test cases and 1,058 interaction turns, where each case specifies a world setting and a multi-turn interaction sequence, covering diverse scenes, styles, subjects, and both first- and third-person perspectives, together with four interaction types, including navigation, subject action, event editing, and perspective switching. For navigation, WBench unifies text, 6-DoF pose, and discrete-action control, enabling evaluation of models with different native input interfaces. Evaluation uses 22 automatic sub-metrics that combine specialist vision models with large multimodal models, and all metrics are validated against human judgments. Across 20 state-of-the-art models, we find that no single model performs strongly across all dimensions. We provide detailed diagnostic insights into the characteristic strengths, weaknesses, and open challenges of each model. Code and data are available at https://github.com/meituan-longcat/WBench.

As personal agents evolve to handle complex, user-centric tasks, static plain-text chat is rapidly becoming a bottleneck. Generative UI emerges as the necessary new interface layer, dynamically synthesizing the right controls, options, and state from the interaction context in real time. We present Macaron-A2UI, a model for Generative UI in personal agents. Our goal is to move beyond text-only interaction by enabling agents to generate natural language together with lightweight, executable UI actions for information collection, preference refinement, confirmation, and multi-goal organization. We build a large-scale Generative UI corpus from heterogeneous dialogue sources, introduce A2UI-Bench for controlled evaluation, and train 30B, 235B and 754B models with parameter-efficient LoRA-based supervised fine-tuning followed by reward-driven reinforcement learning. The best Macaron-A2UI model reaches 75.6 overall on A2UI-Bench without explicit schema hints, surpassing the strongest full-schema frontier baseline. We release the models, benchmark, and evaluation protocol to support future work on Generative UI for personal agents.

Autonomous agents are moving from tools into a layer of social infrastructure: they browse, purchase, deploy software, manage systems, and increasingly interact with one another. As these systems scale, the bottleneck shifts away from raw model capability toward coordination. Agents need to form reliable relationships, organize multi-agent work, exchange value, support an AI economy, and stay safe and accountable under real-world oversight. This paper introduces the Foundation Protocol (FP), a graph-first coordination layer for an emerging human-AI society. FP unifies heterogeneous entities, including agents, tools, resources, humans, institutions, and organizations, and supports native multi-party organization and event-based collaboration. It also provides economic primitives for metering, receipts, and settlement, and treats policy, provenance, and audit as first-class concerns. FP is designed to wrap and bridge existing protocols rather than replace them, enabling incremental adoption while reducing integration and governance overhead. The aim is to keep autonomous agency composable while keeping accountability non-negotiable, so that coordination itself can become shared infrastructure for a human-AI society that is open, pluralistic, and governable.

Sparse-view 3D reconstruction is increasingly addressed with feed-forward splatting networks that predict explicit primitives directly from images. Yet most existing methods remain centered on Gaussian primitives and expose surfaces only indirectly: extracting a usable mesh for downstream simulation, physics reasoning, or embodied interaction still requires expensive post-hoc steps that break the feed-forward promise. This limitation is especially pronounced in pose-free settings, where scene structure and camera parameters must be estimated jointly from sparse observations. We present TriSplat, a feed-forward reconstruction network that represents scenes with oriented triangle primitives and directly exports simulation-ready mesh scenes from a single forward pass. Given input images, the network predicts local 3D point maps, triangle attributes, camera poses, and optional intrinsics. Rather than regressing triangle orientation as an unconstrained latent variable, our approach constructs geometry normals from the predicted point maps, refines them with an image-conditioned normal head, and converts them into stable local frames for triangle parameterization. A mono-normal bootstrap schedule further stabilizes early training, while opacity and blur scheduling progressively sharpens the learned surface representation for direct mesh extraction. Experiments on RealEstate10K and DL3DV show that this representation produces more geometry-faithful reconstructions than Gaussian feed-forward baselines while maintaining competitive novel-view rendering quality. Because the rendering primitives are themselves surface triangles, the output can be directly ingested by physics engines, collision detectors, and standard rendering pipelines without any conversion, making it a practical simulation-ready solution for feed-forward 3D scene reconstruction.

Multimodal modeling represents a vital step from modality-agnostic reasoning toward world modeling. While early approaches predominantly rely on late-fusion that assembles encoders and frozen language backbones with output heads, recent efforts have shifted the paradigm toward native multimodal modeling (NMM) with the intrinsic integration of modalities for superior multimodal performance. Despite its potential, the design space of native architectures remains insufficiently defined. In this paper, we present the community with a formalized roadmap for this transition. Specifically, we formally define the architectural nativity, distinguishing mid-fusion and early-fusion from non-native paradigms. We further organize the existing native models through the lens of input-output duality into three categories: (i) Multi-to-Text for cross-modal comprehension with text-only output; (ii) Multi-to-Target for scenario-oriented generation, e.g., image, audio and video generation, and (iii) Multi-to-Multi for unified modeling with symmetric input-output. We deliver a comprehensive and industrial-grade investigation into the transition toward the definitive NMM framework, where understanding and generation seamlessly coexist within a unified transformer paradigm. We systematically unpack the end-to-end pipeline from industrial perspectives from architectural coordination, massive data curation, to full-stack training recipes, inference & deployment, and the comprehensive evaluation for truly native modeling.

Training large multimodal models (LMMs) via reinforcement learning (RL) to natively invoke video-processing tools (e.g., cropping) has become a promising route to long-video understanding. However, existing native-RL methods dispatch tool calls sequentially (i.e., one per turn): a single wrong crop propagates errors without peer correction, multi-turn tool calls corrupt context, and inference cost scales linearly with the number of turns. We introduce ParaVT, the first multi-agent end-to-end RL-trained framework for Parallel Video Tool calling, dispatching multiple time-window crops in a single turn for cleaner context and better fault tolerance. Yet applying standard RL to ParaVT reveals an obstacle we term the Tool Prior Paradox: the pretrained tool priors that enable tool exploration also destabilize cold-started structural format and expose the skip-tool reward shortcut under temperature sampling. A cross-model contrast on a weaker-prior LMM supports this claim: format stays stable but RL elicits zero tool calls, indicating that prior strength is the shared driver of both format collapse and tool exploration. We propose PARA-GRPO (Parseability-Anchored and Ratio-gAted GRPO), which augments standard RL with two complementary mechanisms: (i) a targeted format reward applied only at the structural-token positions most prone to collapse, and (ii) a per-prompt frame-budget randomization that creates training prompts where calling the tool yields a measurable reward signal over skipping it. Across six long-video understanding benchmarks, ParaVT improves over the Qwen3-VL baseline by +7.9% on average, with PARA-GRPO lifting training-time format compliance from 0.13 to 0.64. As tool capabilities become increasingly internalized in modern LMMs, RL must cooperate with the resulting priors, and ParaVT offers a general recipe for agentic RL. Code, data, and model weights are publicly available.

Efficient attention algorithms are critical to mitigate the quadratic cost of attention in long-context workloads. Prior work utilises block-scaled quantisation techniques on Blackwell GPUs to move attention computation to 4-bit precision to accelerate inference. However, these techniques result in significant quality degradation in long-context settings. We show that the output impact of quantisation error is highly non-uniform and increases with the importance of each query-key interaction, concentrating functionally relevant error in a small number of attention blocks that contain the most important tokens. We propose ThriftAttention, a low-bit attention variant that delivers near-FP16 long-context quality at FP4 inference efficiency. This approach proceeds in two stages. First, a heuristic rapidly selects a small number of important query-key block pairs for FP16 precision. Second, the selected blocks are computed in FP16 and the remaining blocks in FP4, with both paths merged via online softmax into a single output. We demonstrate across long-context benchmarks and model families that by computing only 5% of query-key blocks in FP16, ThriftAttention recovers on average 89.1% of the FP4-to-FP16 performance gap. We show ThriftAttention's advantage grows with sequence length, mitigating the systematic FP4 quality degradation observed at longer contexts. The code is available at https://github.com/joesharratt1229/ThriftAttention.

Deep research agents extend the role of search engines from retrieving keyword-matched pages to synthesizing knowledge, fundamentally changing how humans interact with information. However, frontier systems remain proprietary, while existing open agents often generalize poorly across different task types, leaving unclear how to train a broadly capable deep research agent. We release QUEST, a family of open models (ranging from 2B to 35B) that serve as general-purpose deep research agents designed to handle a wide range of long-horizon search tasks, with strong capabilities in fact seeking, citation grounding, and report synthesis. To build QUEST, we propose an effective training recipe combining mid-training, supervised fine-tuning, and reinforcement learning. Central to this recipe is a curated data synthesis pipeline based on unified rubric trees, which applies to different task types and enables synthesizing training data with verifiable rewards without human annotation. In addition, QUEST incorporates a built-in context management mechanism that enables effective long-horizon reasoning and knowledge synthesis. Using only 8K synthesized tasks, QUEST approaches or even surpasses frontier closed-source agents across eight deep research benchmarks spanning diverse task types, and achieves the best overall performance among recent open-weight agents. We released everything: models, data, and training scripts.

Scientific research is being reshaped by AI systems that move beyond isolated assistance toward longer-horizon workflows spanning literature grounding, hypothesis generation, experimentation, validation, reporting, and revision. This shift marks a transition from task-level AI for science to workflow-level research automation. Yet current systems remain fragmented, differing in autonomy, domain scope, execution environment, validation mechanism, and human oversight, while still struggling with evidence preservation, reproducibility, weak-direction rejection, provenance tracking, cross-domain robustness, and accountable scientific closure. This survey examines these developments through AutoResearch, defined as the developmental spectrum of AI-powered scientific workflow automation. Within it, Vibe Research denotes the human-steered region of prompt-based assistance and human-verified execution, whereas emerging AI-led systems coordinate larger portions of the discovery loop without achieving robust autonomy. We analyze how research systems redistribute control, evidence, execution, validation, and accountability across workflows and organize the field around five workflow conditions: literature and research grounding; hypothesis formation and planning; experimentation and tool use; feedback, validation, and review; and reporting and knowledge communication. We further synthesize AI scientist systems, mixed-initiative co-research frameworks, benchmarks, domain deployments, and open-source infrastructures. Finally, we propose five evaluation dimensions--novelty, validity, impact, reliability, and provenance--and show that AutoResearch autonomy is domain-conditioned, being more credible in structured, executable, and rapidly verifiable settings but limited in embodied, delayed, heterogeneous, ethical, or institutionally accountable contexts.

Multimodal retrieval relies heavily on single-vector retrievers, which compress rich, sequential token sequences into one single global representation. While efficient, they discard fine-grained, local evidence critical for dense retrieval tasks. Multi-vector approaches were introduced as a solution, but they strictly require training and many ignore the necessity of a globally summarizing representation. To address this, we introduce SMART, a framework that unlocks the latent multi-vector capabilities of standard single-vector models. We first demonstrate that standard contrastive training on the pooled embedding implicitly shapes the retrieval geometry of preceding hidden states via gradient flow. By applying direct late-interaction over these frozen hidden states during inference, SMART acts as a plug-and-play upgrade that consistently improves performance across diverse modalities, improving even the state-of-the-art models further on MMEB-V2. We also reveal SMART's superior performance, as simple lightweight post-training not only saves time and compute, but also brings forth further improvement on Visual Document retrieval, allowing a single-vector model to outperform SoTA multi-vector counterparts. Ultimately, SMART offers both a highly efficient inference enhancement and a powerful finetuning technique for multimodal retrieval. We open source our code and weights at https://github.com/HanSolo9682/SMART.

Generating complete digital twins from videos requires precise camera control, global scene coverage, and strict spatial-temporal consistency constraints that remain challenging for perspective video generators due to their limited field of view (FoV). Their narrow FoV forces long or multi-view trajectories, amplifying cross-view inconsistency and temporal drift. We argue that 360° video generation offers a natural solution: panoramic coverage simplifies trajectory design and provides a strong global context for maintaining coherence. We introduce Pantheon360: Taming Digital Twin Generation via 3D-Aware 360° Video Diffusion, a controllable 360° video generation framework that synthesizes high-fidelity videos from sparse 360° inputs. The key idea is an explicit 3D Cache, reconstructed from the input, which serves as a geometric scaffold for any user-defined camera path. This allows the diffusion model to focus on photorealistic texture refinement while the 3D Cache enforces global geometric consistency. Experiments show that Pantheon360 achieves superior visual quality and unmatched geometric coherence, enabling reliable and flexible 360° scene generation for downstream simulation and digital-twin applications.

Existing deep learning-based low-light enhancement methods are typically trained on limited datasets with single enhancement targets, which restricts their generalization ability and controllability in real-world applications. To overcome these limitations, we propose ControlLight, a controllable, consistent, and generalizable framework for low-light enhancement. We first construct a large-scale dataset of real-world degraded images with continuous illumination-strength supervision. To further ensure consistent outputs under different control strengths, we introduce a misalignment-aware weighted flow matching loss that preserves image structure across continuous enhancement strengths. ControlLight allows users to edit real-world degraded low-light images toward satisfactory enhancement results by flexibly controlling the strength while preserving visual consistency and realism. Extensive experiments show that ControlLight achieves state-of-the-art performance against existing low-light enhancement approaches while demonstrating strong continuous controllability and generalization to real-world scenarios.

Reinforcement learning with verifiable rewards (RLVR) has driven breakthroughs in domains such as math, tool-use, and software engineering, yet its extension to computer-use agents (CUAs) has been bottlenecked by the scarcity of scalable training data with deterministic rewards. Constructing such data for CUAs requires consistent task instruction, executable environment, and verifiable reward. However, hand-curated benchmarks achieve high reward fidelity but cover few applications and LLM-as-judge-based datasets scale broadly but lack reliable verification. We present CUA-Gym, a scalable pipeline that co-generates task instructions, environment states, and reward functions. Concretely, a Generator agent constructs the initial and golden environment states, and a separate Discriminator agent writes the reward function from the task specification. An orchestrator agent drives the two through iterative rounds upon execution. Generated tuples then pass a final filter combining LLM majority voting and agent rollouts, ensuring quality beyond the per-task adversarial loop. To address the scarcity of training environments, we further synthesize CUA-Gym-Hub, a broad suite of high-fidelity mock web applications grounded in real-world software-use distributions, expanding the scale of CUA RLVR data by magnitude. Using this pipeline, we construct CUA-Gym, a dataset of 32,112 verified RLVR training tuples grounded in 110 environments. Trained with GSPO on CUA-Gym, our CUA-Gym-A3B and CUA-Gym-A17B achieve 62.1% and 72.6% on OSWorld-Verified, outperforming prior open-source CUAs at comparable scales, with performance scaling smoothly in both data volume and environment diversity. The same checkpoints also improve on the held-out WebArena benchmark, indicating transfer beyond the training environments. We will open-source the full synthesis pipeline, dataset, CUA-Gym-Hub environments, and models.

While AI agents demonstrate remarkable capabilities in reasoning and tool use, they remain fundamentally reactive: they compute responses only after explicit user prompts. This paradigm ignores a critical opportunity: the idle time between interactions is largely wasted, leaving agents unable to prepare for future user needs. To bridge this gap, we introduce ProAct, a proactive agent architecture that leverages idle-time compute to anticipate and fulfill likely upcoming user needs. By analyzing evolving dialogue history together with persistent memory, ProAct predicts upcoming needs and iteratively acquires information, allowing the agent to resolve knowledge gaps and prepare evidence before the user initiates a query.To rigorously evaluate proactive capabilities, we also introduce ProActEval, a comprehensive benchmark comprising 200 scenarios across 40 domains, featuring predictable need chains and diverse user cognitive profiles. Empirical results demonstrate significant advantages over reactive baselines. ProAct accelerates task completion by reducing required turns by 14.8%, decreases user effort by 11.7%, and cuts hallucination rates by 28.1% on ProActEval. Furthermore, MemBench evaluations confirm that ProAct achieves state-of-the-art reflective accuracy, underscoring its sustained and robust performance.

Axios : Sources: Bond Capital is leading a new investment for AI startup Suno, which would value it at ~$5B, up from $2.45B last fall; Suno is expected to raise $250M+ —  Bond Capital is leading a new investment for AI music startup Suno, which would be valued at around $5 billion, Axios Pro has learned.

Lawrence Abrams / BleepingComputer : Charter confirms a data breach after ShinyHunters claimed to steal 40M customer records from Charter's Salesforce instance and threatened to leak the data —  U.S. telecommunications giant Charter Communications has confirmed it suffered a data breach after the ShinyHunters extortion group threatened …

Samantha Subin / CNBC : Micron hit a $1T market value for the first time on Tuesday as shares jumped 19%, driven by demand for its memory chips in the AI race —  Micron topped a $1 trillion market value for the first time on Tuesday as shares popped 19%, driven by insatiable artificial intelligence demand for its memory chips.

Jeremy Kahn / Fortune : London-based Perceptic, which says its end-to-end AI platform for drug development is being used by top pharmaceuticals, raised a $12M seed led by Accel —  A trio of former Palantir executives who helped spearhead that company's Life Sciences practice have founded a startup called Perceptic …

Nilay Patel / The Verge : Q&A with Sundar Pichai about reshaping the information ecosystem with Search changes, putting AI agents in everything, when AI will replace him as CEO, and more —  Today, I'm talking with Google and Alphabet CEO Sundar Pichai, in a conversation we recorded just after the Google I/O developer conference.

Tom Warren / The Verge : Nvidia officially retires its GeForce Control Panel app after 20 years, following the porting of all of its major features to the Nvidia app —  Nvidia has now ported across all of the major Control Panel features to its Nvidia app. … Nvidia announced more than two years ago that it was working …

Ivan Mehta / TechCrunch : Human Archive, which trains robots using first-person video from 1,000+ camera-equipped caps worn by Indian home services workers, raised $8.2M from YC and more —  In the last few years, India's online food delivery market has grown significantly, with both Zomato and Swiggy going public …

Alex Weprin / The Hollywood Reporter : SpaceX filing: X's ad revenue was $1.8B in 2025, $1.7B in 2024, and $2.3B in 2023, below Twitter's $4B in 2021; X and Grok now have 6.3M active paid subscribers —  SpaceX's IPO filing reveals ad revenue for X is still below what it was when Musk acquired Twitter, but it is growing again.

Leslie Josephs / CNBC : American Airlines picks SpaceX's Starlink for in-flight Wi-Fi on more than 500 planes; SpaceX already has contracts with United Airlines, Southwest, and others —  American Airlines plans to outfit more than 500 of its narrow-body aircraft with Starlink, handing another win to Elon Musk's SpaceX unit …

Ian King / Bloomberg : Sources: Qualcomm reached a deal with ByteDance to supply millions of ASICs for AI data centers to support AI agents in the Doubao chatbot; QCOM jumps 4.48% —  Qualcomm Inc. reached a deal with TikTok owner ByteDance Ltd. to supply chips for artificial intelligence data centers …

Financial Times : Filing: Monzo reports its “refer a friend” payouts grew 40% YoY to £29.5M for the 12 months ending March 2026, as part of a broader £143M marketing spend —  London-based fintech records ‘staggering’ increase in spending on referral fees

Koh Ewe / BBC : China executes a man called Xu Yao for killing Yoozoo Games founder Lin Qi in 2020; Lin reportedly sidelined Xu, who helped land the 3 Body Problem Netflix deal —  Shanghai First Intermediate People's Court  —  Chinese authorities have executed a man for murdering his associate, billionaire gaming tycoon Lin Qi.

Kate Knibbs / Wired : How AI startups like Altur are using chatbots to help automate debt collection; YC incubated six debt collection and settlement startups in the past six years —  There's a mad dash to automate the world's most hated calls.  Have an unpaid bill?  You'll hear from an AI debt collector sometime soon.

Jonathan Vanian / CNBC : Dropbox founder Drew Houston is stepping down as CEO after 19 years to become executive chairman, replaced by Ashraf Alkarmi, who is SVP and GM of Dropbox Core —  Drew Houston founded Dropbox nearly two decades ago out at age 24, eventually becoming a household name in Silicon Valley …

Max Buondonno / The Shortcut : Google Fitbit Air review: slim, comfortable, and stylish, robust tracking, seven-day battery life, and cheaper than Whoop, but can only be worn on the wrist —  🏆 Rating: 4/5  —  ✅ Pros  — 📐 Slim design that fades into the background  — ⌚️ Comfortable and stylish band options

2021 年美国德州通过法案禁止孕妇在妊娠约 6 周后堕胎。2022 年美国最高法院在 Dobbs v. Jackson Women’s Health Organization 一案中裁决宪法未赋予公民堕胎权，因此推翻了 1973 年的 Roe v. Wade 案。截至 2026 年初美国有 13 个州全面禁止堕胎，7 个州禁止孕妇妊娠 22 周后堕胎。严格堕胎禁令被认为会增加妊娠相关死亡率。发表在《American Journal of Public Health》期刊上的一项研究调查了严格堕胎禁令对孕妇健康的影响。结果显示，在 14 个严格禁止堕胎和禁止妊娠 6 周后堕胎的州，妊娠相关死亡比预期高 9.2%。

Meta 在 2021 年开源了缓存引擎 CacheLib，该项目旨在利用非易失性存储器作为缓存去扩展服务，以抵消不断上涨的 DRAM 成本。该项目在 2024 年 6 月之后就停止了更新，但在 2026 年 5 月 25 日 Meta 再次释出了更新——而今天由于 AI 热 DRAM 价格相比 2021 年几乎是天价。

科学家首次记录了一次非凡的鲸类迁徙壮举，证实两头座头鲸在澳大利亚东部和巴西的繁殖地之间，穿越了超过 1.4 万公里的海洋。研究人员通过对比数万张座头鲸尾鳍的图像来辨认这些鲸。每头鲸的尾鳍都有独特的斑纹，这使得研究人员能长期追踪并识别个体。2007 年，一头座头鲸在澳大利亚昆士兰州的赫维湾首次被拍到。2013年，它再次出现在同一海域，随后于 2019 年现身巴西圣保罗附近。这些繁殖地之间的最短直线距离约为1.42万公里。第二头座头鲸更令人惊叹。研究人员于2003年首次在巴西阿布洛霍斯礁群——该国主要的座头鲸繁殖地，拍摄到了它的身影。当时它正与由9头成年鲸组成的活跃群体一起游弋。22年后的2025年9月，同一头鲸被发现在澳大利亚赫维湾独自游弋。两次目击地之间的距离达 1.51 万公里，这创下了单头座头鲸已知最远迁徙距离的新纪录。研究基于19283张高质量的鲸照片，这些照片拍摄于1984年至2025年间，采集自澳大利亚东部和拉丁美洲。这些图像既来自专业研究人员，也来自通过全球鲸追踪平台“Happywhale”参与的公民科学家。

英国皇家医学院学会在递交给政府的咨询意见书中表示，社交媒体的使用与吸烟一样对年轻人的健康构成威胁。医生在接诊年轻患者时，应例行询问他们的屏幕时间和社交媒体使用情况。英国政府正在考虑的一项措施是禁止 16 岁以下儿童使用社交媒体，类似澳大利亚的做法。其它可能采取的限制包括宵禁，或禁用自动播放和无限滚动等功能。儿童精神科医生 Emily Sehmer 认为过度使用社媒的危害远甚于吸烟，因为儿童只需几秒钟就会接触到有害内容。

Uber 高管表示 AI 上支出并没有带来相应的回报。Uber COO Andrew Macdonald 上周六接受采访时表示愈来愈难以证明最大化 AI 词元花的钱是合理的。而在上个月的一次采访中 Uber CTO Praveen Neppalli Naga 告诉 The Information，该公司已经用完了 2026 年的 Claude Code 预算。Macdonald 称，通过与工程主管的交流，他认识到更高的 AI 词元使用量并没有转变为消费者功能的相应增加。他说 AI 带来的权衡成本愈来愈难以证明支出是合理的。

JAXA、早稻田大学、东京大学和庆应义塾大学的工程师团队成功完成了为五马赫高超音速飞机设计的冲压式发动机的地面燃烧试验。冲压发动机利用了发动机的前向运动来压缩空气，不使用带有可旋转叶片的压气机，它无法在空速为零的时候产生推力，需要先加速到超音速。在测试中，一架实验飞机被安装在 JAXA 角田宇宙中心的风洞中，模拟约 25 公里高空的环境条件。在五马赫的飞行速度下，机头和前缘周围的空气温度会超过 1000 摄氏度，为应对高温，工程师设计了一套先进的热防护系统，使飞机内部温度保持在接近正常工作温度的范围内，保证机载航空电子设备和控制电子设备的正常运行。JAXA 接下来计划将实验飞行器搭载在探空火箭上尝试实际飞行，它的目标是到 2040 年代实现商业高超音速客运服务。

欧洲 ESA 和日本 JAXA 合作的水星探索项目 BepiColombo 以意大利数学家 Giuseppe Colombo 的名字命名，探测器于 2018 年 10 月发射，原计划在六次飞掠水星之后于 2025 年 12 月进入水星轨道。但第四次飞掠水星前推进器出现故障，地面任务规划人员不得不修订时间表。JAXA 通过其社交媒体账号宣布了最新的日期：BepiColombo 计划于 11 月 21 日进入水星轨道。BepiColombo 包含三个组件：ESA 的水星转移模块和水星行星轨道器，以及 JAXA 的水星磁层轨道器。JAXA 的轨道器分离时间定在 12 月 10 日。BepiColombo 是人类第三次水星探测任务，前两次是 1973 年的 Mariner 10 和 2004 年的 Messenger。水星是太阳系最小密度最高的行星，由于温度非常高，ESA 的轨道器安装了上百公斤的隔热材料。

加州年龄验证法律的修正案将豁免大部分 Linux 发行版和自由开源软件。年龄验证法律要求操作系统提供商在设置询问用户年龄。该法案的修改版 AB-1856 缩小了适用的操作系统提供商和应用程序的范围：(2) “操作系统提供商”不包括在许可条款允许接收方复制、重新分发和修改该软件的情况下，分发操作系统或应用程序的个人或实体。(2) “应用程序”不包括其本身并未作为独立可执行应用程序、通过受监管的应用程序商店向消费者提供的软件组件。Valve 的 SteamOS 平台仍然受到影响，因为它的 Steam 客户端是受监管的应用商店。

中亚的冰川是生活在下游地区的数百万人的重要水源。一项新研究发现 2025 年中亚经历了创纪录的冰川损失。冰川加速消融可能会在短期内增加融水，但最终由于冰量的减少融水也会减少。研究人员利用对天山和帕米尔高原 16 座冰川的实地观测数据，结合模型，估算出中亚冰川在一年内损失了约 30 立方千米的冰，相当于该地区冰川总体积的近 2%。这一结果是异常温暖的春夏季气温以及降雪频率的大幅下降造成的。16 座冰川有 9 座经历了有史以来最严重的冰川质量损失，帕米尔高原西部和天山山脉西部的冰川消融最为严重，部分冰川在一年内损失了 2%-4% 的总冰量。64% 的冰川经历了自 1991 年以来最严重的冰川质量损失。研究人员警告由于全球暖化，这种情况可能成为常态。

用户通过社交媒体报告，摩托罗拉手机预装的一个应用 Smart Feed 在更新之后开始劫持亚马逊应用植入联盟营销推广码获取佣金。非常奇怪的是，推广码 sramz-kff-008-20 指向的是一名时尚博主“@kirasfashionfinds”，也就是佣金给的不是 Smart Feed 而是这位博主。暂时不清楚究竟发生了什么。受影响的用户可通过禁用 Smart Feed 关闭推广码，方法是：Settings > Apps > 搜索定位到 “Smart Feed” > Disable。

教宗良十四世颁布了其首道通谕，呼吁世人不可用人工智能来作恶，切莫把人工智能当成「掌控、排斥或死亡的工具」。 教会长期支持核裁军，称这是「为人类大家庭和平与尊严的服务」。同样地，人工智能今天也不可用于作恶，这就「如同核能那样，必须用来为所有的人和公共福祉效劳」。「关于科技的决定绝对是与良心和责任密不可分的」。「和平不只是没有战争，更是正义伸张。然而，当科技削弱我们的批判意识时，和平本身就会陷入危险。无论如何，光是解除武装仍有所不足，我们还必须进行建设。」

欧洲刑警组织披露，他们黑进了被网络犯罪分子使用的 VPN 服务“First VPN”，访问了用户数据库，识别了数千用户身份。First VPN 的网站已经显示被执法部门扣押的信息，它过去曾在俄语网络犯罪论坛上打广告，宣称能隐藏用户的 IP 地址，加密所有通信，不记录任何日志。它还声称将拒绝与司法机关合作，其服务不受任何司法管辖，且不会存储任何用户数据。First VPN 的活动始于 2014 年，在 27 个国家/地区提供了 32 个出口节点服务器。至少有 25 个勒索软件组织利用了其基础设施进行网络侦察和入侵。警方搜查了该服务管理员在乌克兰的住所，拆除了 33 台服务器。

对英伟达、AMD、Google 和亚马逊四家公司的 AI 芯片的分析显示，HBM 内存芯片成本占到了 AI 芯片组件成本的三分之二（63%），逻辑芯片占 13%，先进封装占 15%，辅助组件占 9% 。四家公司在 HBM 上的支出从 2024 年的约 120 亿美元增至 2025 年的 320 亿美元，增速远超其它芯片组件。随着内存芯片供应持续紧张且价格上涨，HBM 在 2026 年的市场份额可能会进一步扩大。超大规模数据中心运营商在其资本支出预期中已经预见到这一点：微软 2026 财年 1900 亿美元的资本支出预期中，约有 250 亿美元来自组件价格上涨；Meta 将其 2026 年资本支出预期上调了 100 亿美元，理由同样是组件价格上涨。

过去几个月惠普笔记本电脑用户通过论坛等报告在更新 BIOS 之后设备出现了问题，包括设备无法启动、风扇噪音异常以及蓝屏死机等等。一名移动工作站 ZBook Ultra G1a 的用户称更新 BIOS 之后设备在启动过程中卡住。受影响的产品包括 ZBook Ultra G1a，存在问题的 BIOS 版本号 01.04.03 和 01.04.05；EliteBook X G1a，存在问题的 BIOS 版本号 01.03.11 和 01.05.00。惠普表示它正对此展开调查，建议受影响的用户联系其技术支持团队。这不是第一次惠普设备因为存在问题的 BIOS 更新而导致设备故障。

俄罗斯政府已推迟对使用 VPN 的移动互联网用户收费的计划。俄罗斯数字发展部在三月表示将打击 VPN 的使用。它最初要求移动网络运营商从 5 月 1 日起对每月国际数据流量超 15GB 的用户收费。但由于追踪 VPN 使用和计费方面存在困难，该期限已推迟至 6 月 1 日。该收费计划可能会再次被推迟，可能会在 9 月底国家杜马和地方选举之后实施。原因是一个功能完整的国际流量支付系统需要三到四个月才能建成。在这项政策推行前，俄罗斯的移动互联网频繁发生中断事件。