OrangeBot.AI Digest — 2026-06-25

Safe agent spend from the terminal

CLI that turns plain-English into real browser tests

Build a web app by talking through a meeting

Web browser automation for AI agents

Let AI agents run your next deal, fundraise or data room

Scale across AI models without scaling your bill

The always-on computer for your coding agents

Live autonomous AI DJ that digs, mixes & explains 24/7

Let AI agents see cross repo and maintain session memory.

Live AI teammate for every tough sales conversation

Voice-to-text for macOS with a fully on-device option

One AI that knows your entire company and acts on it

Your second brain for product management

Turn followers into clients.

Native project planning app, now on Mac & with an MCP server

Memory for large language model (LLM) agents has rapidly evolved from simple retrieval-augmented mechanisms into a data management system that supports persistent information storage, retrieval, update, consolidation, and dynamic lifecycle governance throughout agent execution. Despite this evolution, existing evaluations still benchmark agent memory mainly through end-to-end task success metrics (e.g., F1, BLEU), while treating the underlying system as a monolithic black box. As a result, critical system-level concerns, including operational costs, architectural trade-offs across memory modules, and robustness under dynamic knowledge updates, remain insufficiently explored. In this paper, we present a systematic experimental study of agent memory from a data management perspective. We propose an analytical framework that decomposes agent memory into four core modules: memory representation and storage, extraction, retrieval and routing, and maintenance. Under this framework, we evaluate 12 representative memory systems and two reference baselines across five benchmark workloads spanning 11 datasets. Our extensive end-to-end evaluation shows that no single architecture dominates across all scenarios; instead, effectiveness depends heavily on how well the memory structure aligns with the workload bottleneck. Furthermore, through fine-grained ablation studies, we quantify their individual effects on representation fidelity, retrieval precision, update correctness, and long-horizon stability. Finally, we reveal cost-performance trade-offs under realistic workloads, showing localized maintenance is more cost-efficient than global reorganization. Based on these findings, we identify promising directions towards building truly agent-native memory systems. The code is publicly available at https://github.com/OpenDataBox/MemoryData.

Open domain subject-driven text-to-video (S2V) generation has drawn significant interest in academia and industry. Open domain S2V mainly involves two scenarios: in-domain, which requires retaining the reference subject features as much as possible, and cross-domain, which preserves the intrinsic features of the subject while allowing subject-irrelevant properties to vary flexibly according to the text prompt. Existing methods primarily focus on maximizing subject fidelity in in-domain scenarios, which limits their editability and adaptability in cross-domain scenarios, such as novel styles, semantic combinations, or domain attributes. In this study, we propose that an ideal S2V method should flexibly shuttle between different domains, achieving strong performance in both in-domain and cross-domain scenarios. To this end, we propose DomainShuttle, which could achieve high fidelity and generative flexibility for open domain video personalization. Specifically, we introduce Domain-MoT, which decouples videos and reference features and introduces the domain-aware AdaLN for domain-specific modeling of reference images. We then introduce the Video-Reference DualRoPE scheme, which places reference image tokens and video tokens in separate RoPE spaces to enable precise subject-level spatial modeling, and Cross-Pair Consistent Loss, which aims to extract intrinsic subject features unaffected by irrelevant features. Extensive experiments demonstrate that DomainShuttle achieves significant performance improvements over existing methods, exhibiting high subject fidelity and generative flexibility across diverse open domain application scenarios.

We present Wan-Streamer, a native-streaming, end-to-end interactive foundation model designed from the ground up for real-time, low-latency, full-duplex audio-visual interaction. Wan-Streamer seamlessly models language, audio, and video as both input and output within a single Transformer, where the sequence is represented as interleaved visual, audio, and text input tokens together with visual, audio, and text output tokens, coordinated by block-causal attention for incremental streaming. Unlike cascaded interactive systems that rely on separate VAD, ASR, language, TTS, audio-driven animation, or video-generation modules, Wan-Streamer does not rely on external language, speech, avatar, or video-generation modules: perception, reasoning, generation, response timing, turn management, and cross-modal synchronization are learned jointly within one unified model, reducing pipeline latency and error accumulation. To support natural audio-visual responsiveness, we redesign the entire stack around streamability, including causal encoders, causal decoders, block-causal attention, and low-latency multimodal token scheduling, enabling streaming units as short as 160 ms at 25 fps. Wan-Streamer achieves approximately 200 ms model-side response latency and approximately 550 ms total interaction latency when combined with 350 ms bidirectional network latency, supporting sub-second duplex audio-visual communication. These results position Wan-Streamer as a unified, end-to-end, multimodal interactive foundation model for low-latency streaming interaction.

Real-world photography requires capture-time guidance for both camera framing and subject pose. Yet existing aesthetic cropping benchmarks mainly evaluate post-hoc crop prediction and overlook subject-side recommendations, leaving the capture-time guidance capabilities of multimodal large language models (MLLMs) underexplored. To address this gap, we introduce CaptureGuide-Bench, a benchmark with two complementary tasks: photographer-side composition decision and refinement, and subject-side scene-conditioned pose recommendation. Our evaluation reveals limitations: general-purpose MLLMs can make composition decisions but lack precise refinement localization, while specialized aesthetic cropping models localize crops effectively but are limited to refinement; neither provides actionable pose guidance. To support model development, we further construct CaptureGuide-Dataset, comprising 130K samples with textual rationales and structured visual annotations, and develop ShutterMuse, a unified MLLM trained with supervised and reinforcement fine-tuning. Experiments on CaptureGuide-Bench show that ShutterMuse achieves the best overall photographer-side performance among evaluated baselines and competitive subject-side pose recommendation with substantially lower inference cost, demonstrating the potential of MLLMs as interactive assistants for photography during image capture.

Modern large language models are predominantly trained with autoregressive factorization and causal attention. We present iLLaDA, an 8B masked diffusion language model trained from scratch with fully bidirectional attention. iLLaDA keeps the masked diffusion objective throughout pre-training and supervised fine-tuning (SFT), scaling pre-training to 12T tokens and fine-tuning on a 25B-token instruction corpus for 12 epochs. We further use variable-length generation for efficiency and introduce confidence-based scoring for multiple-choice evaluation. Compared with LLaDA, iLLaDA improves broadly across general, mathematical, and code benchmarks; for example, iLLaDA-Base improves by 21.6 points on BBH and 14.9 points on ARC-Challenge, while iLLaDA-Instruct improves by 14.5 points on MATH and 16.5 points on HumanEval. Despite its non-autoregressive training, iLLaDA also remains competitive with Qwen2.5 7B on several benchmarks. These results show that fully bidirectional diffusion training from scratch is a competitive path toward strong language models. Model weights and codes: https://github.com/ML-GSAI/LLaDA.

While Large Language Models (LLMs) have substantially advanced text-to-code synthesis, many real programming tasks specify intent through visual artifacts such as screenshots, charts, vector drawings, videos, and interactive states. These tasks require models to connect visual perception to executable programs, because correctness depends not only on syntax but also on layout, data semantics, interaction behavior, and domain-specific constraints that apply after execution. This survey examines Multimodal Code Intelligence, covering systems that generate, edit, refine, or reason with code under visually grounded inputs and outputs. We first formulate the field by the role that code plays in each task, distinguishing code as a rendered artifact, an editable symbolic structure, a scientific representation, an intermediate reasoning trace, or an executable policy or tool interface. We then organize benchmarks and methods into four domains: Graphical User Interface, Scientific Visualization, Structured Graphics, and Frontier Tasks and Frameworks. This taxonomy connects mature artifact-generation problems to emerging agentic and unified settings and allows us to compare how different tasks treat evidence of correctness. Looking ahead, we argue that future research may benefit from four verification-centered directions. Multi-signal validation can combine complementary evidence of correctness, multi-state verification can test behavior across execution trajectories, cross-task transfer testing can probe reusable visual-code skills, and verifiable agent traces can reveal whether agent actions are grounded in visual evidence. Together, these directions may move this field from single-output imitation toward evidence-grounded executable systems. An ongoing project and resources are available on https://github.com/xjywhu/Awesome-Multimodal-LLM-for-Code{GitHub}.

Synthesizing a novel-view video from a monocular reference video along a target camera trajectory requires both geometric consistency and motion fidelity with respect to the reference video. Existing methods based on explicit 3D representations are limited by the accuracy of off-the-shelf reconstruction modules, which often produce inaccurate geometry for dynamic objects in monocular videos. In contrast, camera-conditioning-only methods can achieve high visual quality but often struggle to preserve geometric and motion consistency. In this work, we introduce MVTrack4Gen (Multi-View point Tracking for Novel-View Generation), a motion-aware training framework that leverages multi-view point tracking as an additional geometric and motion supervision signal for camera-conditioning-only novel-view video diffusion models. Our key finding is that specific attention layers encode strong correspondence cues, where query features attend to key features at geometrically corresponding locations across views and over time, and the misalignment of these correspondences causes motion inconsistency. Based on this observation, we route these features into an auxiliary multi-view tracking head and jointly train the diffusion model with a point-tracking objective. By explicitly strengthening these motion-aware correspondences, MVTrack4Gen improves existing models to better follow the motion in the reference view and maintain cross-view geometric consistency. Across diverse benchmarks, our method achieves state-of-the-art geometric consistency and competitive camera accuracy.

Generating a coherent multi-shot video requires structured cross-shot memory. Subject appearance, scene context, and speaker identity must persist across cuts. Existing approaches either train end-to-end over fixed-length sequences and cannot scale, generate shot-by-shot with memory banks that grow linearly, or orchestrate pretrained generators under an LLM planner without a multi-shot-aware backbone. We present UnityShots, a memory-driven multi-shot audio-video generation system built on LTX-2.3, trained on annotated cinematic and music-video shots. The video stream maintains two fixed-size slots, a long-term memory (LTM) slot anchored to the opening shot and a short-term memory (STM) slot holding the immediately preceding tail, both updated at every cut by a boundary-conditioned gate that fuses visual cut probability and beat-tracker signals. The audio stream injects a reference speaker token at every shot to preserve vocal timbre without a sliding audio bank. A discrete cut-type prior, learned through AdaLN, becomes an inference-time control knob over transition strength. We release a benchmark of 200 multi-cultural multi-shot sequences spanning six ethnic regions and ten or more languages, with per-shot reference identities, reference audio, and per-boundary transition labels. Evaluated across I2V, T2V, and R2V conditioning modes, UnityShots leads open-source baselines on every cross-shot coherence metric and matches the strongest closed-source system on the multi-shot axes.

Fine-grained visual reasoning requires multimodal large language models (MLLMs) to identify task-relevant visual evidence and ground their reasoning in local image regions. Existing agentic methods typically rely on reinforcement learning with verifiable rewards or supervised fine-tuning on large-scale annotated reasoning traces, leading to costly exploration, hand-designed verification rules, or heavy dependence on textual supervision. A natural way to avoid such external answer labels is to learn from trajectories sampled by the student itself, which points to On-Policy Distillation (OPD). To understand what OPD can and cannot provide for visual reasoning, we revisit it as negative-free stop-gradient alignment. This perspective shows that, although OPD provides effective token-level correction, its ceiling is constrained by the absence of trajectory-level discrimination. Motivated by these observations, we propose V-Zero, an answer-label-free framework for visual reasoning with contrastive evidence gating. V-Zero uses no annotated textual answer labels; instead, during training it pairs a question-relevant regional crop with a negative visual view to evaluate student-sampled trajectories and gate dense token-level distillation. Experiments on multiple visual reasoning benchmarks show that V-Zero consistently improves fine-grained visual reasoning while preserving strong generalization. Notably, V-Zero is more than 5times faster than previous supervised fine-tuning methods and more than 10times faster than reinforcement learning baselines. Code and dataset will be released at https://github.com/eVI-group-SCU/V-Zero

Autoregressive video diffusion with causal diffusion transformers has emerged as a major paradigm for real-time streaming video generation and action-conditioned interactive world models. In this work, we extend rCM, an advanced diffusion distillation framework, to autoregressive video diffusion. The core philosophy of rCM lies in the complementarity between forward and reverse divergences, represented by consistency models (CMs) and distribution matching distillation (DMD), respectively, in diffusion distillation. This philosophy naturally carries over to the autoregressive setting, where teacher-forcing (TF) provides an offline, forward-divergence causal training paradigm, while self-forcing (SF) corresponds to an on-policy, reverse-divergence refinement. Our contributions are: (1) through extensive experiments, we show that teacher-forcing CM is currently the best complement to self-forcing DMD as an initialization strategy (2) we present the first implementation of teacher-forcing-based continuous-time CMs (e.g., sCM/MeanFlow) for autoregressive video diffusion, enabled by our custom-mask FlashAttention-2 JVP kernel, achieving 10times faster convergence compared to discrete-time CMs (dCMs) (3) we introduce Causal-rCM, a leading, unified, and scalable algorithm-infrastructure open recipe for diffusion distillation and causal training (4) we achieve state-of-the-art streaming video generation performance in both frame-wise and chunk-wise settings, using only synthetic data for training. Notably, our distilled 2-step causal Wan2.1-1.3B model achieves a VBench-T2V score of 84.63 with only 1 or 2 sampling steps. We further apply Causal-rCM to Cosmos 3, an advanced omnimodal world foundation model for physical AI with action-conditioned generation capability, enabling an interactive world model.

We present EBench, a simulation benchmark that diagnoses generalist mobile manipulation policies beyond a single success-rate scalar. EBench comprises 26 diverse and challenging manipulation tasks annotated along 5 capability dimensions and 4 generalization dimensions. We evaluate state-of-the-art generalist manipulation models including π_0, π_{0.5}, XVLA, and InternVLA-A1, and reveal that models with near success rates exhibit strikingly different capability profiles: π_{0.5} achieves the highest test success rate and the best train--test retention, whereas InternVLA-A1 dominates mobile manipulation but collapses on dexterous tasks, and XVLA exhibits strengths on a disjoint set of atomic skills compared to other policies. Beyond capability profiling, EBench analyzes the generalization ability from 4 representative perspectives, identifying the impact of different distribution shift factors. The results reveal strengths and weaknesses of models behind an overall score. We hope this benchmark offers a broad set of diagnostic signals to guide iteration on generalist manipulation models.

Unified multi-modal large language models (MLLMs) have achieved strong text-to-image generation quality, but still struggle with structure-aware prompt following, where object counts, spatial relations, attribute bindings, and coarse layouts must be preserved. We attribute this limitation in part to the entanglement of structural planning and appearance rendering within a single conditioning stream. To address this issue, we propose Implicit Visual Chain-of-Thought (IV-CoT), a latent visual reasoning framework for query-conditioned image generation. IV-CoT decomposes the visual conditioning queries into a structural-to-semantic cascade, where structural queries first form a latent visual plan and semantic queries then render appearance conditioned on this plan. To guide the structural queries, we introduce training-only sketch supervision, which encourages them to capture structure from sketches without requiring sketch extraction or intermediate decoding at inference time. IV-CoT performs implicit CoT reasoning in a single forward pass and achieves superior results on GenEval and T2I-CompBench. Visualizations and analyses demonstrate that the learned structural and semantic queries play complementary roles in structure-aware generation.

The Hitchhiker's Guide to Agentic AI is a comprehensive practitioner's reference for building autonomous AI systems. The book covers the full stack from first principles to production deployment, organized around a central thesis: building great agentic systems requires understanding every layer of the pipeline, not just one. The book opens with the LLM substrate -- transformer architecture, GPU systems, training and fine-tuning (SFT,LoRA, MoE), model compression, and inference optimization -- treated as essential foundations rather than the primary focus. It then develops the alignment and reasoning layer: reinforcement learning from human feedback (RLHF), PPO, DPO and its variants, GRPO, reward modeling, and RL for large reasoning models including chain-of-thought and test-time scaling. The second half is devoted to agentic AI proper. Topics include agentic training and trajectory-based RL, retrieval-augmented generation (RAG and Agentic RAG), memory systems (in-context, external, episodic, and semantic), agent harness design and context management, and a taxonomy of agent design patterns. Inter-agent coordination is covered in depth: the Model Context Protocol (MCP), agent skills and tool use, the Agent-to-Agent (A2A) communication protocol, and multi-agent architectures spanning centralized, decentralized, and hierarchical topologies. The book concludes with agent development frameworks, agentic UI design, evaluation methodology for agentic tasks, and production deployment. Each chapter pairs rigorous theoretical foundations with implementation guidance, code examples, and references to the primary literature.

We introduce Autodata, a general method that enables AI agents to act as data scientists who build high quality training and evaluation data. We show how to train (meta-optimize) such a data scientist agent, so that it learns to create even stronger data. We describe the overall formulation, and a specific practical implementation, Agentic Self-Instruct. We conduct experiments on computer science research tasks, legal reasoning tasks and reasoning with mathematical objects, where we obtain improved results compared to classical synthetic dataset creation methods. Further, meta-optimizing the data scientist agent itself delivers an even larger performance uplift. Agentic data creation provides a way to convert increased inference compute into higher quality model training. Overall, we believe this direction has the potential to change the way we build AI data.

Chain-of-Thought (CoT) has become a standard method for improving reasoning capabilities in large language models (LLMs) by eliciting step-by-step thinking, but its effectiveness in multimodal tasks remains unclear. In this paper, we aim to systematically investigate the key question: What can multimodal Chain-of-Thought reasoning do, and where and why does it fall short? To this end, we evaluate 12 multimodal tasks across perception and reasoning categories using both 14 non-reasoning models and 8 reasoning models. Our analysis reveals several important findings: (1) CoT is not a free lunch and should be used selectively depending on the specific requirements of each task. For perception tasks, CoT can lead to undesirable side effects, such as reduced performance in visual grounding and object counting. In contrast, it proves effective for reasoning tasks involving mathematical, scientific, and multi-image reasoning; (2) Compared to original models, existing open-source multimodal reasoning models often yield only marginal overall improvements, possibly due to an overemphasis on mathematical reasoning at the expense of broader capabilities; (3) Visual reasoning remains a key bottleneck for current multimodal CoT, as models exhibit a Look Light, Think Heavy pattern where verbal reflection rises and falls during reasoning, whereas visual reflection consistently diminishes. These findings suggest that while multimodal CoT handles verbal reflection relatively well, it lacks the ability to maintain deep visual introspection throughout the reasoning process.

Marina Temkin / TechCrunch : Patronus AI, which builds simulated digital environments for evaluating AI agents, raised a $50M Series B led by Greenfield, bringing its total funding to $70M —  AI agents are becoming more sophisticated.  They are evolving from answering questions to autonomously executing multi-step complex tasks.

Om Malik / On my Om : Om Malik, a longtime technology writer, founder of Gigaom, and a partner at True Ventures, died on Wednesday at 59 —  Om Malik passed away on June 24, 2026, at Stanford Hospital after a long health journey with his heart.  He was surrounded by family and friends.

Ann Gehan / The Information : Source: Google is asking publishers that test new AI features in Google News to grant it broad rights to their content, including to potentially train AI models —  Google is tightening the screws on news publishers.  —  In recent months, the tech giant has been pitching publishers …

Samantha Subin / CNBC : Chipmaker Onsemi agrees to buy Synaptics in a nearly $7B all-stock deal expected to close in the middle of 2027; ON drops 9%+ and SYNA jumps 11%+ after hours —  On Semiconductor has agreed to buy Synaptics in a nearly $7 billion all-stock deal to bolster its push into physical artificial intelligence technology.

Alan Neuhauser / Axios : Source: Muon Space, which operates satellite networks for climate and national security monitoring, is raising $250M; it has raised ~$181M since its 2021 launch —  Muon Space, which builds and operates satellite networks, is raising $250 million, Axios Pro has learned.

New York Times : Sources: OpenAI leans toward holding off its IPO until 2027 after warnings that Sam Altman's desired $1T valuation may not be met in current market conditions —  The A.I. company's advisers are pushing its chief executive, Sam Altman, to move slowly after SpaceX's stock has been volatile …

The Information : Sources: Sam Altman told staff the US government asked OpenAI to stagger the release of GPT 5.6 over security concerns, approving “access customer by customer” —  For AI companies on the verge of releasing cutting edge new AI models, there's a new normal in the wake …

MacKenzie Sigalos / CNBC : Apple stock closed down 6.15% on Thursday after the company raised some product prices, its worst fall since April 2025 —  Apple on Thursday announced price hikes on MacBooks and iPads, its first formal move to pass higher memory and storage costs on to consumers after CEO Tim Cook said increases had become unavoidable.

Olivier Acuna / CoinDesk : Story Protocol, a blockchain-based IP ownership network that raised $140M, rebrands as Data Foundation to build an on-chain registry for AI training data —  Palo Alto-based blockchain startup Story Protocol is rebranding as DATA Foundation and shifting its focus entirely to AI training infrastructure …

James Attwood / Bloomberg : The US awards $250M in CHIPS funds to a startup co-founded by billionaire mining magnate Robert Friedland to develop silicon-carbide chips and pulsed-power tech —  I-Pulse Inc. will receive $250 million in US funding for semiconductor and pulsed-power development as the venture co-founded …

Zac Bowden / Windows Central : Microsoft quietly extends the Extended Security Updates program for Windows 10 consumers by a year, letting eligible users get updates through October 12, 2027 —  Windows 10's ESU program has been quietly extended by an extra year, now ending on October 12, 2027 instead of October 2026.

Ram Iyer / TechCrunch : Netris, which helps neoclouds automate network configuration to bring their data centers online faster, raised a $15M Series A from a16z —  The AI boom has encouraged everyone and their uncle to launch a data center business.  But spinning up a data center isn't easy.

Will Canny / CoinDesk : Sources: Kraken is in talks to acquire a 15% stake in DeFi protocol Aave at a $385M valuation, investing 35,000 ETH in return for 250,000 AAVE tokens —  Crypto exchange Kraken, part of Payward Inc., is in talks to acquire a 15% stake in decentralized finance (DeFi) protocol Aave at a $385 million valuation …

Zac Hall / 9to5Mac : Notion plans to shut down its Gmail client Notion Mail on September 22 and go “all in” on AI agents to run inboxes, saying 50%+ of users do not open the inbox —  Last year, Notion expanded its productivity suite to include Notion Mail, an AI-powered email client.

Kenneth Shepard / Kotaku : Microsoft says the price of Xbox consoles will increase on August 1 by $100 for 512GB models and $150 for 1TB models, the third price increase since 2025 —  Microsoft is increasing the prices of Xbox Series consoles once more and introducing a ‘Buy Now, Pay Later’ option

密码管理器 LastPass 再次披露了用户数据泄漏事故，这一次是它的外部合作伙伴 Klue 导致的，黑客访问了客户信息和支持案例数据。LastPass 称，被访问的数据包括客户姓名、电话号码、电子邮件地址和实际地址，以及支持案例数据和销售相关数据。它表示在获悉数据泄漏之后，它立即撤回了员工对 Klue 的访问，轮换了暴露的 API 令牌，通知了执法部门。LastPass 警告客户对钓鱼攻击或社交工程攻击提高警惕，公布了与攻击者相关的 IP 地址和电邮域名。

在苹果 CEO 库克提前透风数天之后，苹果产品全系列涨价，涨幅少则 50 美元多则上千美元。即使是苹果也无法再自己承担高昂的内存和存储器成本。 苹果在一份声明中表示，“我们从未见过一个组件价格以如此之快、如此之大的幅度上涨。迄今为止，我们一直在尽力为客户抵挡这些涨价，但现在我们已经到了不得不开始提高部分产品价格的地步，包括今天 iPad 和 Mac 的涨价。我们知道这不是一个好消息，我们正在不遗余力地寻找解决方案。”

生殖专家曾认为，女性绝经后，卵巢会像阑尾一样变得无用。在对 50-75 岁女性的卵巢进行检查时，研究人员发现该器官的细胞会随着年龄增长产生不同的蛋白质。为了更深入研究卵巢的年龄相关变化，研究人员转向了实验小鼠。尽管小鼠不会出现雌激素急剧下降等人类更年期特有特征，但这些动物在 2 年生命周期的后期，卵巢功能也会停止。研究人员分别从年轻小鼠、处于生殖期末期的小鼠以及“绝经”后小鼠体内摘取了卵巢。对每只动物，他们对其中一侧卵巢的 RNA 进行了测序，以测量基因表达情况。对另一侧卵巢，他们对组织进行了显微镜下视觉分析，以识别不同的细胞群，并测量纤维化的发展程度，纤维化是指随着年龄增长自然发生的硬化组织堆积现象。但对“绝经”后卵巢的分析显示，其中各类免疫细胞的水平均高于年轻小鼠的典型水平。此外，老年小鼠的卵巢中，编码各种促炎化合物的基因活性更高，这些免疫分子可能被分泌到血液中并随血液流向身体其他部位。尚不清楚衰老的卵巢究竟是真正发挥着免疫信号传导的作用，还是仅仅是免疫细胞的意外聚集地。这一发现或许有助于解释，为何女性尽管寿命更长，但随着年龄增长，健康状况往往不如男性。绝经后的卵巢可能会分泌某些分子，导致女性在更年期出现慢性炎症。

镉不是植物生长的必要元素，但其通过土壤—水稻—食物链进入人体长期摄入后，会引发肾功能损伤、癌症、骨质疏松等严重健康问题。OsNramp5 是水稻中负责从根部往茎部运输镉的关键转运蛋白，但也同时负责锰离子等植物生长必需的金属离子的运输，敲除 OsNramp5 可以有效降低镉的运输，但也会造成其他必要金属元素的缺乏，使水稻大幅减产。根据发表在 PNAS 期刊上的研究，中国科学院遗传与发育生物学研究所等通过碱基替换技术，靶向编辑水稻负责吸收镉元素的核心转运基因 OsNramp5，创制了优异人工等位变异，发现了特异降低镉吸收而不影响锰等其他关键金属离子吸收的新机制，解决了低镉与高产难以兼顾的难题，为镉污染农田安全生产主粮提供了可落地的育种新方案。

OpenAI 宣布了首款自研芯片 Jalapeño，由 OpenAI 与博通公司合作设计和制造，专门用于 AI 推理。OpenAI 没有披露技术方面的细节，只是称初步测试显示每瓦性能显著优于目前最先进的同类产品。OpenAI 与博通是在去年 10 月正式宣布合作，OpenAI 声称利用其模型加速了芯片的设计。自研 AI 芯片旨在减少对英伟达的依赖，Google 和亚马逊也都开发了自研芯片。

英国维基百科员工率先寻求成立工会。维基媒体基金会英国员工于 6 月 24 日星期三致函管理层，请求由 Communication Workers Union(CWU)下辖分支 United Tech and Allied Workers (UTAW) 代表他们的权利。员工呼吁维基基金会作为这家全球非营利机构的实际管理者，履行其领导层最近作出的公开承诺，即保障员工组织和组建工会的权利。逾千名维基志愿者和社区成员签署了请愿书声援这些员工。英国是仅次于美国的维基媒体基金会第二大员工来源国。

微软更新了 Surface 购买指南，声称 8GB 内存对 Windows 11 足够日常使用了，如浏览、视频串流、作业和生产力应用。它同时表示 16GB 或以上的内存才能解锁 Copilot+ PC 功能。由于内存短缺且价格翻了数倍，PC 厂商不得不开始提供 8GB 内存的设备，但 8GB 内存对 Windows 11 而言非常勉强，而过去两年微软的宣传是 16GB 内存是获得良好 Windows 11 体验的必要条件。作为主要 AI 基础设施提供商，微软当然也是造成今天这一局面的罪魁祸首之一了。

美国白宫今年五月宣布其白宫应用将自动下载到政府配发手机上。该应用无法卸载，即使政府雇员尝试卸载，应用也会很快重新安装。美国农业部、国务院和劳工部员工匿名接受采访时表示，这款应用出现在手机上时让他们感到不安，有人试图删除它，但失败了。“我把它删了，测试下，结果它立刻又出现了，”一位美国农业部雇员说。白宫应用内有一个按钮，允许用户“给特朗普总统发短消息”，点击后会自动弹出一个写着“史上最伟大总统”的文本框。应用的社交部分可看到来自白宫 X 账号推文、特朗普 Truth Social 账号发布的帖子，以及官方账号在 TikTok 和 Instagram 等平台上分享的视频。新闻部分包含了白宫新闻稿、简报和情况说明书，以及来自 Fox, Breitbart, Reuters, The New York Post 等媒体的精选文章，这些内容要么对本届政府政策大加赞扬，要么攻击民主党。一位政府雇员说这是赤裸裸的宣传。

我们习以为常的图形 UI 中的每一个小细节，无论多么微小，都是由某个人在某个时间点想出来的。举例来说：拼写错误的单词下方的小红色波浪线。这种设计已成为每个文本编辑字段司空见惯的元素，以至于无人特意去思考它。然而它确实是由某个人发明的，微软资深程序员 Raymond Chen 说，这个人是 Tony Krueger。早期的 Word 版本中，拼写检查功能需要用户手动调用，然后等待程序查找所有可能拼写错误的单词，逐一向用户展示，由用户决定如何处理每一个错误。Word 引入了自动拼写检查功能，在用户空闲时运行拼写检查，当用户点击拼写检查按钮时，结果已准备就绪。然而自动拼写检查仍然是一个阻塞操作。很多用户选择关闭它，因为它总是会在你想做其它事情如保存并退出时突然决定“现在是检查文档拼写的好时机”，迫使你等待拼写检查完成。Tony 让拼写检查器变得更不显眼，不会干扰用户的当前工作。当它发现问题时，不会触发拼写检查，而是立即在可能拼写错误的单词下画上红色波浪线，后来在可能语法错误的单词下画上绿色波浪线。

对 LG 和三星智能电视应用的扫描发现，6038 款电视应用中有 2058 款嵌入了住宅代理 SDK，也就是会出售用户的家用 IP 作为代理服务使用。智能电视是理想的代理主机，它基本上一直处于插入电源状态，同时接入了家用 WIFI，但不像 PC 没人会去检查其可疑后台活动。电视应用上的广告可能会让用户不满，但默默运行的住宅代理则能在最小化用户不满的同时给运营商带来收入。但住宅代理会有滥用的风险，Kimwolf 僵尸网络就滥用了住宅代理进行传播和扩散。

Anthropic 指控阿里巴巴的 Qwen AI 实验室非法蒸馏其 AI 模型。Anthropic 在给美国议员的信中称，阿里巴巴的上述行动发生在今年 4 月 22 日至 6 月 5 日期间，通过近 2.5 万个欺诈账户与 Claude 进行了超过 2880 万次交互。这封日期为 6 月 10 日的信在一场有关 AI 的听证会前发送给美国参议院银行委员会主席蒂姆·斯科特和资深成员伊丽莎白·沃伦。

科学家在南非 Wonderwerk 洞穴发现了新证据，表明人类祖先在 107-179 万年前就开始使用火，这是已知最早的人类用火记录。研究人员在洞穴深处约 30 米处发现了反复用火的痕迹，这些地点远离自然野火可能影响的范围，因此表明早期人类有意将自然产生的火带入洞穴并持续燃烧。早期人类不能随意生火，他们很可能是从闪电引发的火或草原野火收集火源。

根据市场分析公司 Omdia 的数据，中国一季度 PC 出货量下滑 2%，平板电脑下滑 5%。PC 出货量降至 890 万台，平板电脑出货量降至 830 万台。笔记本电脑（含移动工作站）出货量同比下降 19%，而台式机（含台式工作站）出货量同比增长 41%，分别达到 530 万台和 360 万台。Omdia 称市场疲软的原因是组件成本上涨导致设备价格上涨，以及消费者补贴力度减弱。Omdia 预测 2026 年全年 PC 出货量将下降 14% 至 3600 万台，平板电脑出货量预计将下降 11% 至 3200 万台。最主要 PC 制造商包括联想、华为、苹果、软通动力和惠普。

随着屏幕在幼儿生活中几乎无处不在，一项研究调查了其对学习表现的影响。研究跟踪了 1-8 岁的儿童，发现屏幕观看时间更长与 9 岁时较差的学习表现以及 10.5 岁时较弱的工作记忆存在关联。研究结果表明，屏幕接触的时机可能与屏幕使用的总时长同样重要。WHO 和美国儿科学会建议幼儿在 18–24 个月前不要接触屏幕，2-5 岁儿童每天使用屏幕时间不超过 1 小时。但很多幼儿都超过了这些限制。最新研究追踪了 502 名儿童从婴儿期到童年中期的发育过程，发现在特定发育阶段屏幕观看时间较长的儿童，后期学业表现较差，工作记忆较弱。这种关联在婴儿期和学龄初期最为显著，表明这些阶段可能是认知发展的特别敏感窗口期。在整个童年期屏幕接触总量较高的儿童，学业表现也通常较差。研究结果表明，屏幕使用的时机可能与总暴露量同样重要。研究结果支持“越少越好”的原则，即儿童的屏幕时间越少越好。

本周英国、法国、意大利和西班牙都发布了红色高温预警，欧洲正经历五月以来第二波热浪。全球气温比工业化前时期——1850-1900 年——的水平高出约 1.4C，而根据欧盟哥白尼气候变化服务中心的数据，欧洲气温比工业化前水平高出约 2.4C。全球平均气温的持续上升主要是由于燃烧石油、天然气和煤炭产生的温室气体排放，但由于多种因素的共同作用，不同地区的升温幅度不同。陆地升温速度快于海洋，因为水可以吸收更多热量并通过蒸发冷却。哥白尼气候变化服务中心称，大气环流的变化导致欧洲夏季热浪更频繁强度更大。另一个主要原因是地理位置，欧洲与北极相连，北极气温比工业化前水平高出 3.2C。北极地区气温上升的部分原因是反照率。明亮的冰雪会将大部分太阳热量反射回太空，但冰雪融化会露出颜色较深吸收热量的陆地。欧洲冬季降雪频繁的地区，积雪覆盖面积正在减少，露出了深色的陆地。