OrangeBot.AI Digest — 2026-08-06

The fastest way to send anything to your iPhone

A practice app for learning to play music with your hands.

Turn any website into LLM-ready Markdown

Free, open-source AI bug-fixing agent

Build the AI operating system for your company

Track your AI spend and connect it to business outcomes

Screen recording as prompts

Make your product feed visible to AI shopping agents

Stop your AI from guessing brand logos

Publish sites, newsletters and social posts with AI agents

An RSVP reader for books, PDFs, articles & webpages

delivered tomorrow morning at a random time between 8 -10 AM

Define safety at runtime for text and images

The easiest way to access frontier AI models.

Bring any agent to Slack, Teams, and more.

Long-horizon search agents must make multiple sequential actions (steps) to search, retrieve, verify, and integrate evidence to reach a final answer. However, existing methods for training these agents typically treat all steps within a trajectory uniformly during both supervised fine-tuning (SFT) and reinforcement learning (RL), failing to distinguish useful actions from erroneous or redundant ones. In this paper, we propose Answer-Backtracked Credit Assignment (ABC), a fine-grained credit assignment framework for training long-horizon search agents by converting sparse trajectory-level outcomes into dense step-level supervision that rewards useful actions (even in failed trajectories) while suppressing erroneous or redundant actions. Specifically, given a potentially obscure query and its corresponding ground-truth answer, ABC first performs Answer-Backtracked Clue Recovery, which traces back from the answer to recover intermediate clues required to solve the question. It then applies Clue-Anchored Step Scoring to evaluate each search step against these clues, converting sparse binary outcome supervision into dense step-level rewards. Based on these rewards, we develop ABC-SFT, which reweights the loss of each turn, and ABC-GRPO, which uses the step-level scores as rewards in GRPO. Building on this framework, we train ABSeeker based on Qwen3.5-4B with only 8.5k examples. ABSeeker achieves 37.3% on BrowseComp and 39.1% on BrowseComp-ZH. With context management, the scores further improve to 55.3% and 52.9%, respectively, significantly outperforming same-scale (4B) agents and even matching the performance of larger ones (approximately 30B). These results demonstrate the effectiveness of answer-backtracked step-level credit assignment for training long-horizon search agents.

Text-to-image (T2I) models can produce visually compelling images, yet they remain limited on open-world tasks that require complex semantic understanding, multi-step reasoning, and the integration of external world knowledge. Existing efforts introduce agent capabilities into image generation, but they either prescribe a fixed workflow or place only a subset of the open-world image generation process under agent control. Consequently, reasoning, tool invocation, and image generation are not coordinated by a single policy. We propose ToolArtist, a fully agentic image generation model obtained by post-training a Unified Multimodal Model (UMM). ToolArtist dynamically orchestrates reasoning, external tool use, and native image generation within one unified policy. During Supervised Fine-Tuning (SFT), we equip a teacher agent with search tools alongside an image-generation tool. We then convert the collected trajectories into a UMM compatible format, where the image-generation tool is concealed while the resulting generated images are retained. During Reinforcement Learning (RL), we develop an agentic RL infrastructure for UMMs and introduce Reason-Act-Draw GRPO (RAD-GRPO), which uses complementary intent and quality rewards to jointly optimize the model. Experiments show that placing the entire open-world image-generation process under an agent policy consistently outperforms approaches with fixed pipelines or only partially agent-controlled components. We release the training data and the complete post-training infrastructure.

Vision offers a critical axis for advancing foundation models, driving a shift towards natively unified multimodal pretraining. Despite this momentum, the design space and the fundamental mechanisms of how modalities interact during unified training remain underexplored. We provide empirical clarity through a systematic exploration of multimodal pretraining. Our controlled experiments on both synthetic and large-scale real-world datasets yield four key insights into the physics of multimodal pretraining: (i) Knowledge Flow: We disentangle how language, visual understanding, and visual generation transfer knowledge across modalities, revealing distinct patterns of influence and asymmetry; (ii) Synergy vs. Competition: We show that data "complexity" largely determines whether modalities are synergistic, identify architectural choices that promote synergy: such as shared attention and normalization with modality-specific feed-forward layers, and find that these behaviors generalize across different visual tokenizer designs; (iii) Early Unification: Unifying modalities from the very early stages and training them jointly is shown to be more effective than late alignment or sequential training. This process uncovers a vision laziness phenomenon, where delayed integration leads models to rely on language priors; (iv) Recipes: We derive efficient pretraining recipes that achieve strong generative performance using only 5% of the compute budget. These core findings are subsequently validated at scale by training multiple 13.5B MoE models on 2T tokens. We hope this study provides a principled foundation for understanding and scaling multimodal pretraining.

Personalized LLMs with persistent memory are increasingly deployed, yet the faithfulness of their user models remains unexamined. We study over-inference (OI): the phenomenon where LLMs fabricate user attributes beyond what evidence supports. We introduce MirageBench, comprising 150 personas balanced across stereotypical, counter-stereotypical, and neutral profiles, 6 personalization tasks spanning an ``imagination gradient'', a four-way faithfulness taxonomy operationalized by an independent judge (validated against a blind human annotator on 400 claims: Cohen's kappa = 0.863 four-class, kappa = 0.900 binary), and a leaderboard of 12 models across 7 families on 143616 judged claims. We find that over-inference is pervasive: every one of the 12 models over-infers 35%--49% of its claims (cross-model mean 41.6%; claim-weighted 41.8%), with no model in this evaluation escaping it. Most strikingly, we surface a Self-Monitoring Inversion: at the model-selection level, models' self-assessed OI is negatively rank-correlated with their judge-measured OI (rho = -0.60, p = 0.044; exploratory, wide bootstrap CI [-0.90, +0.06], n = 12). The models that report the least over-inference tend to be flagged as fabricating the most, so self-reported confidence is a misleading signal for comparing models, even though within a single model self-audit still ranks that model's own claims moderately well (AUROC 0.58--0.83). We further show that OI is task-dependent (27%--59%) and that, in a multi-turn pilot, inferred attributes accumulate approximately linearly with little revision. MirageBench positions external verification, rather than model self-report, as a more reliable foundation for trustworthy personalization.

LLM agents are increasingly applied to open-ended everyday requests that span work, study, and life. These tasks are long-horizon, cross-environment, and multimodal, forcing the agent to preserve goals and constraints across many steps while navigating heterogeneous tools and attachments. While prior work has addressed individual failure modes such as goals drift, states loss, and context overflow, whether a single harness can manage them jointly and remain effective across backends has received less study. We present OneDayAgent, a long-horizon harness for autonomous agents. OneDayAgent turns an open-ended request into a managed execution process that decomposes tasks into bounded subtasks, maintains execution memory under context pressure, and verifies and repairs the final deliverable. We evaluate OneDayAgent on AgentIF-OneDay across 104 tasks. With the GLM-5.2 backend, OneDayAgent sets a new state of the art with an overall score of 0.821. The same harness runs across five backend LLMs from three model families, indicating the harness generalizes across backends without tuning, even as different models induce distinct execution styles under the same workflow.

Agent self-evolution updates an agent's persistent state from prior experience and reuses it to solve related tasks more effectively. Evaluating self-evolution is difficult: existing benchmarks provide limited coverage of economically valuable task domains, do not always design training and test tasks such that test-time gains can be attributed to training experience, and remain vulnerable to data contamination. We present GDPevo, an evolution-native benchmark grounded in GDP-related enterprise workflows, together with the fully automated data pipeline that generates it. Its core mechanism, rule hybridization, decomposes each enterprise workflow into atomic business rules, distributes subsets of these rules across training tasks, and recombines them in held-out test tasks so that test-time gains are attributable. GDPevo spans CRM, ERP, finance, healthcare, legal, and data-centric workflows. Its V1 release contains 120 tasks in 12 groups, with five training and five held-out test tasks per group. Full automation enables the pipeline to expand the suite to 240 tasks in 24 groups (V2) within two days, providing a practical response to contamination. Using GDPevo, we evaluate four agents, each comprising a harness and a model, under four supervision types. Self-evolution consistently improves held-out accuracy by up to 16.44 percentage points. But the best evolved agents remain far below the fully informed oracle ceiling of 91.6%, indicating that the self-evolution ability of current agents remains far from fully realized. We publicly release the pipeline, benchmark, and full evaluation results at https://github.com/Prism-Shadow/GDPevo.

Long-horizon reasoning in recent LLMs demands that the model switch between distinct skills inside a reasoning chain, such as first doing a math derivation, then using the result to plan a schedule. We call such problems cross-skill long-horizon tasks: multi-step tasks whose steps require different reasoning skills and depend on earlier outputs. Existing benchmarks often evaluate individual skills, lacking a principled way to measure how well a model switches between skills. We address this gap from both the evaluation and training sides. We introduce Skill Entropy, a measure of the difficulty of switching from one skill to another. We then propose Skill^2-Bench, a benchmark of cross-skill long-horizon tasks built over 558 skills across 9 verifiable and open-ended domains. Each task is assigned a task-level skill-entropy score and grouped into three difficulty levels. Evaluating 8 frontier and 4 open-source models on Skill^2-Bench reveals a skill-switching gap: accuracy decreases on higher-entropy tasks. We then turn skill entropy from a benchmark scale into a training signal. We propose Skill-Entropy RL, an RL framework where the model predicts not only the answer at each step but also the skill used to produce it. The reward combines step-level correctness with a skill-entropy reward that measures the alignment between the model-predicted skill sequence and the gold skill sequence. On Qwen3-4B-Instruct and Qwen3-1.7B, Skill-Entropy RL improves the Skill^2-Bench score from 34.4% to 68.4% and from 14.6% to 40.1%, respectively, outperforming competitive baselines. The same pipeline can be applied to off-the-shelf training data such as OpenR1-Math, indicating that skill entropy is a reusable training signal. Code available at: https://github.com/Gen-Verse/Skill-Entropy-RL

On-Policy distillation (OPD) transfers teacher capabilities by supervising student-sampled trajectories with dense token-level teacher signals. Recent selective OPD methods improve this process by prioritizing signals that are confident, informative, or learnable. However, the assumptions overlook a fundamental failure mode of language models: their token-level judgments can be driven by input-agnostic language priors, formatting conventions, or stereotyped reasoning templates rather than task-specific evidence. We refer to such optimization-relevant but weakly input-grounded supervision as spurious signals in OPD, which may produce large gradients while contributing little task-improving direction. To mitigate this issue, we propose SA-OPD, a Spurious-Signal-Aware On-Policy Distillation framework that identifies and filters misleading token-level supervision based on input-groundedness and optimization impact. SA-OPD introduces a lightweight input-groundedness proxy estimating whether a token-level distillation signal truly depends on the input. It then filters only tokens that simultaneously exhibit low input-groundedness and extreme distillation divergence, thereby removing high-impact spurious updates and achieving fine-grained OPD optimization. Extensive experiments on both large language model (LLM) and vision-language model (VLM) settings demonstrate that SA-OPD consistently outperforms Vanilla OPD and competitive selective methods. These results establish input-groundedness as a key dimension for OPD supervision selection and offer a simple, effective strategy for mitigating spurious updates.

Learning generalizable robot manipulation policies requires large-scale and diverse demonstration data. Egocentric human manipulation videos offer rich scene and task diversity, and prior work has shown that retargeting and rendering such videos into robot-format data can yield effective per-task policies at small scale. However, whether this approach can provide pretraining benefits for vision-language-action models at scale remains unexplored. We present Ego2Robot, a scalable pipeline that converts egocentric human manipulation videos into robot training data through action retargeting, robot-arm visual synthesis, and multi-level quality curation. Ego2Robot supports both curated datasets and in-the-wild videos, producing 18,561 hours of robot training data spanning 15 robot morphologies, making it the largest ego-to-robot dataset to date. To evaluate generalization, we extend RoboTwin2.0 with disentangled perturbation axes covering visual appearance, scene layout, embodiment morphology, and task semantics. Experiments show that joint pretraining on Ego2Robot-synthesized and robot data consistently improves out-of-distribution generalization across multiple perturbation types, with benefits validated on real-robot deployment. Project page: https://www-ye.github.io/ego2robot_blog/

We introduce NOLLI, a procedurally generated English-Korean puzzle benchmark designed to diagnose where Korean performance gaps arise. It comprises 15 puzzle types (25 tasks; 7,500 items), with every instance seed-regenerable, verified to have a unique solution, and scored deterministically. Rather than equating harder with bigger, we calibrate difficulty behaviorally, tuning each generator until a fixed reference model lands in target accuracy bands. Its three-level design combines matched direct translations, script adaptations over Hangul jamo (sub-syllabic letters), and Korean-only tasks grounded in Korean culture or orthography. We evaluate 15 frontier, open-weight, and Korean-developed models; among the 12 above a 3% overall-accuracy floor, matched English-Korean accuracy is statistically equivalent within a +/- 10 pp margin (TOST), suggesting little cost from presentation language alone. Writing-system-intensive tasks show sharper gaps: Korean Cipher falls behind English by up to 68.7 pp, whereas Cryptarithmetic over the same jamo shows no systematic penalty, and Jamo Composition accuracy predicts Korean Cipher accuracy. These contrasts are diagnostic rather than causal, consistent with difficulty in multi-step sub-syllabic execution. Korean-only tasks separate rule-application deficits, which vary in sign, from a Kinship deficit positive in all 12. Finally, a salient size measure fails to grow from Easy to Hard in 7 of 15 types, making structural size an unreliable proxy for empirical difficulty.

While instruction-based video editing has advanced rapidly, real-world videos contain tightly coupled audio and visual signals, and editing one modality often requires coordinated changes in the other. Existing benchmarks primarily evaluate visual transformations on silent clips or isolated audio editing, leaving complex audio-visual editing and cross-modal consistency underexplored. We introduce AVE-Compass, a comprehensive benchmark with 145 curated source videos, 196 audio-visually coupled editing instructions, and 2,688 fine-grained checklist items. It evaluates Instruction Following, Fidelity Preserving, Realism, and Editing Intent through checklist-based MLLM judging and a dedicated realism rubric, complemented by automated cross-modal, video, and audio metrics. Extensive evaluation shows that state-of-the-art models still struggle to execute cross-modal instructions while preserving non-target content. We further propose AVE-Agent, a modular agent framework that decomposes complex instructions into dependent subtasks and iteratively improves editing results through self-reflection and evaluator feedback. AVE-Agent improves instruction execution, Fidelity Preserving, and audio-visual alignment in joint editing while maintaining competitive perceptual quality.

This technical report presents K-EXAONE 2.0, an open-weight multilingual foundation model developed by LG AI Research as a step in our effort toward global frontier-scale foundation models. Rather than training from scratch, we upcycle K-EXAONE and expand its architecture, yielding a Mixture-of-Experts (MoE) model with 750B total parameters and approximately 37B activated per token---more than three times the capacity of its predecessor. K-EXAONE 2.0 supports context lengths of up to 256K tokens and expands multilingual coverage from six to ten languages. Its training pipeline combines continual pre-training, difficulty-focused mid-training, and post-training to strengthen reasoning, agentic coding, multilingual capability, and safety grounded in Korean sociocultural contexts. Across nine evaluation categories selected to reflect the conditions of practical use, K-EXAONE 2.0 improves over K-EXAONE and remains competitive with open-weight models, showing its largest gains in agentic coding and long-context understanding and its clearest strengths in long-context retrieval and safety. Released under the Apache 2.0 license, K-EXAONE 2.0 enables the wider AI ecosystem to evaluate, deploy, adapt, and build upon it, while marking the beginning---rather than the endpoint---of our challenge toward the global frontier.

Reinforcement learning with verifiable rewards (RLVR) has become a standard paradigm for post-training large language models (LLMs). While Group Relative Policy Optimization (GRPO) is widely adopted, it suffers from sparse reward signals and loses gradients entirely when all responses within a group receive identical rewards. On-policy distillation (OPD) offers a natural remedy by providing dense, token-level supervision from a teacher model. However, naively combining GRPO with OPD leads to degraded performance, due to three underlying causes: not all samples benefit from distillation; fitting too quickly to the teacher undermines the exploratory capacity of RL; and OPD's advantages are asymmetric, suppressing most tokens. To address these challenges, we propose RSTG (Recovering Learning Signals via Adaptive Teacher Guidance), which applies distillation selectively and precisely where it matters most. At the sample level, OPD is restricted to negative zero-variance prompts with each sample weighted by the teacher's confidence score. At the token level, distillation targets only tokens with high student entropy or large teacher-student divergence. We further augment training with SFT on correct trajectories generated by the teacher model, injecting positive gradient signals where RL yields none. Experiments demonstrate that RSTG substantially outperforms naive GRPO+OPD by +4.02% on math and +3.05% on code.

Memory-augmented VLM agents act on persistent spatial knowledge, yet that knowledge silently goes stale as the environment changes. We ask what happens when an agent must reconcile a confident memory claim with a contradicting observation, and whether current models can catch the conflict before it becomes a safety-relevant mistake. Using a dynamic FrozenLake testbed, we pair a staleness-detection task with a downstream navigation task across three closed-source models and three open-weight VLMs under both text and image inputs (1,800 detection runs, and 12,000 text-mode navigation episodes over four LLM navigators at a shared 50-seed scale). Three findings emerge. First, text solvability does not imply visual grounding: models that flag stale entries reliably from text nonetheless span vision F1 from 0.887 down to 0.067 on the identical grids, and the weakest keeps making fluent, confident decisions that ignore the image. Second, consuming stale memory without an audit is a safety liability: in our primary GPT-4o setting, an agent that trusts raw memory dies more than twice as often as the same agent given no memory at all. Third, auditing helps but does not close the gap: a transparent read-time filter removes much of the safety cost in text mode, yet even oracle stale labels bring no further significant gain on the current grid size, and when visual auditing is unreliable, filtering yields no consistent benefit. Together these results frame spatial-memory staleness as a safety failure mode and isolate reliable visual grounding and action selection under memory--observation conflict as the central open challenges for memory-augmented agents.

Despite the remarkable recent progress of video world models, social interaction between users and the characters within these worlds remains unsupported. To fill this gap, we present HelloWorld, a video world model that enables social interaction with in-world characters. With a single button press, users can prompt the on-screen character to respond toward the camera, e.g., turning to the viewer, waving, nodding, or speaking a short greeting. To make these interactions natural, we propose a self-distillation pipeline that finetunes the video generation model on data synthesized by itself. Each synthesized clip contains both social interactions and camera motion, allowing the model to learn camera-pose conditioning without degrading interaction quality. At inference, we further introduce a training-free module that determines when the interaction occurs. Upon a button press, it modulates the cross-attention masks of the DiT so that the interaction-related text prompt attends only to the frames within the press window, temporally localizing the character's response. We further build HelloWorldBench, a 400-sample benchmark with three social interaction metrics alongside three conventional metrics, for evaluation. Experiments demonstrate that HelloWorld surpasses a variety of baselines in interaction quality, while maintaining state-of-the-art picture aesthetics and camera-pose following. Project page: https://github.com/AlayaLab/HelloWorld

The Information : Sources: Canva slashed revenue growth forecast as heavy use of new AI features drove up costs and slowed their rollout, while more Canva users turned to ChatGPT —  Executives at Canva, the design software app with more than 200 million users, had high hopes for a rollout of AI tools that marked …

Angela Palumbo / Barron's Online : Cloudflare reports Q2 revenue up 36% YoY to $696.1M, above $665M est., and forecasts Q3 revenue above estimates; NET jumps 17%+ after hours —  Choose Barron's as a preferred source of financial news  —  Key Points  —  Cloudflare stock was climbing Thursday night after the company reported better …

Koyena Das / Reuters : Instacart reports Q2 revenue up 14% YoY to $1.04B, GTV up 14% to $10.35B, above est., forecasts Q3 GTV and core profit above est.; CART jumps 8%+ after hours —  Instacart (CART.O) forecast third-quarter gross transaction value and core profit above analysts' estimates on Thursday …

Todd Spangler / Variety : Roku reports Q2 revenue up 22% YoY to $1.35B vs. $1.3B est., net income of $164.2M, ad revenue up 25% to $673M, and subscription revenue up 26% to $548M —  Roku kept its streaming engines blazing in the second quarter of 2026 with earnings that blew past Wall Street estimates.

Kelly Cloonan / Wall Street Journal : Lyft reports Q2 revenue up 16% YoY to $1.84B, vs. $1.81B est., rides up 12% to 262.4M, and expects Q3 gross bookings growth to slow to 15%-19% —  The company's gross bookings rose 23% to $5.5 billion in the latest quarter  —  Lyft's second-quarter bookings rose, boosted by overall demand, particularly for its premium rides.

Natalie Lung / Bloomberg : Airbnb reports Q2 revenue up 17% YoY to $3.6B, Nights and Seats Booked up 10% to 148.3M, above est., lifts 2026 revenue forecast; ABNB jumps 10%+ after hours —  Airbnb Inc. boosted its annual revenue forecast for a second time this year after it saw robust global travel demand, particularly in the US and Europe.

Tobias Mann / The Register : AMD acquires Toronto-based Taalas, which integrates model weights directly into silicon to boost inference performance, for an undisclosed sum —  Early tech demos show model-specific integrated circuits churning out up to 17,000 tokens a second  —  In AMD's latest bid to upset Nvidia's dominance …

Mark Gurman / Bloomberg : Sources: OpenAI's new device, slated for 2027, is a hockey puck-sized smart speaker with moving parts that help give it personality, likely costing $300+ —  A highly anticipated new device from OpenAI will have a unique look, complete with moving parts that help give it personality …

Lara O'Reilly / Business Insider : Gravity, which places text-based ads within AI chatbots, raised a $30.5M Series A co-led by Lightspeed and Committed, taking its total funding to $38.5M —  Zach Oldham, cofounder of the adtech startup Gravity, thinks AI ads will eventually become the world's largest advertising channel.

Anna Tong / Forbes : Docs: US data labeling companies, like Surge AI and Mercor, that sell training datasets to US AI labs and the government are also selling them to Chinese labs —  The same Silicon Valley startups supplying OpenAI, Anthropic and the U.S. government are also selling AI training datasets …

Brody Ford / Bloomberg : Figma CEO Dylan Field forfeits ~$46M in stock awards to revive investor confidence amid AI disruption fears; FIG is down ~77% from its peak more than a year ago —  Video Player is loading.  —  Unmute  —  Current Time 0:00 Loaded: 17.59% Playback Rate  — captions off, selected  — English

Carl Zimmer / New York Times : Scientists trained AI on genetic sequences to design viruses not found in nature, yielding 16 viable viruses that can infect bacteria but don't threaten humans —  Scientists trained artificial intelligence on libraries of DNA and then asked the model to create recipes for viral genomes.

Julia Hornstein / The Information : Sources: Panthalassa, which aims to power data centers in the ocean using energy generated by waves, is raising $225M at a ~$2B valuation, up from $1B in May —  AI data centers need boatloads of power, a need that's set off a flood of investment into startups harnessing nuclear or solar energy to run the servers.

Bailey Lipschultz / Bloomberg : Sources: Nscale told potential backers it generated $100M+ in Q2 revenue, up from ~$37M in Q1, and has ~$51B in contracts, ahead of a possible September US IPO —  Nscale is telling prospective investors it has about $51 billion of total contracted revenue ahead of a US initial public offering …

Terrence O'Brien / The Verge : Suno says it will adopt new audio watermarking tech, fingerprinting tech, transparency tools, and a new download policy to limit the spread of spammy AI music —  The company is changing its download policy and rolling out new watermarking technology.

世界各地的工业采砂活动正以远超地球自然补充的速度开采河床沉积物。世界正在耗尽河沙。研究显示全世界对河沙和砾石的需求是不可持续的。越南目前的采砂速度约为自然沉积物补充速度的 11.8 倍。越南在 2015-2022 年间开采了约 3.66 亿立方米的沙子，导致河床平均下沉 0.48 米。如此大的变化会对水流方式、河流系统的健康以及河流两岸的结构产生重大影响。越南湄公河段在 2018-2020 年间，逾 1800 栋房屋因河岸侵蚀而倒塌。如果越南湄公河目前的采砂速度持续下去，到 2035 年河沙将枯竭。河沙因其颗粒能提供合适的摩擦力，且一般不含盐分，因此在建筑领域广泛使用，它占到了混凝土体积的七成以上。

NASA 未来有意将宇航员送往火星，但火星宇航员如何应对有毒的火星沙尘？NASA 今年早些时候成立了 Martian Dust Limit Working Group，评估已知和未知的火星沙尘对人类宇航员的毒理学危害，最终为人类在火星上设定一个允许的暴露极限值。几十年前的阿波罗宇航员就发现处理月尘很麻烦。月尘会粘附在任何接触到的物体上，会侵蚀宇航服，导致设备和仪器过热，破坏宇航服的密封性，甚至刺激宇航员的眼睛和肺部。火星沙尘不像月尘那样尖锐会割伤人，但火星沙尘有月尘没有的化学成分和矿物质，如对人类有害的高氯酸盐。火星是完全不同的世界，火星沙尘比月尘对健康的危害要大得多。

Reddit 的新版本因为推荐算法和广告等原因而导致用户体验不佳，老用户仍然偏爱旧版本的 Old Reddit(old.reddit.com)。但根据最新的官方博客，Old Reddit 的未来岌岌可危。Reddit 官方声称内容抓取者利用了 Old Reddit 无需登陆的便利而恶意抓取其内容，它将逐步改变 Old Reddit。短期内用户仍然能继续访问 Old Reddit，但将需要登陆。对于所谓的“改变”，Reddit 官方没有给出更具体的说明，只是表示在“探索不同的方案”。Old Reddit 最终可能会消失。

科学家在青藏高原发现的开花植物灯架虎耳草被确认为一个新的食虫植物谱系，以确凿证据实证存在食肉植物类型，也证实进化论奠基人达尔文 150 多年前提出的植物物种可能具有食肉性的预测。研究结果表明，灯架虎耳草能够吸引、捕获并消化昆虫，并从中吸收氮元素。1875 年达尔文曾推测，作为一类通常生长在高山环境中的被子植物，虎耳草属的某些物种可能具有食肉性，因为它们拥有能够捕获昆虫的粘性腺毛。此前一直缺乏确凿证据来支持这一假说。在本项研究中，研究团队对生长在中国青藏高原-横断山脉高山地区的灯架虎耳草进行全面深入研究，实地观察和现有植物样本的调查显示，在 45 个被调查的标本中，有 43 个的腺毛上附着有昆虫猎物，成熟植株平均共捕获 71 只昆虫(主要附着在这些腺毛上)。随后研究团队通过荧光标记法检测到了磷酸酶活性(一种常见于食虫植物的消化酶)，这表明该植物可能具有消化猎物的能力。接下来，研究团队将标记有稳定氮同位素的果蝇喂给灯架虎耳草，以分析该植物能否从猎物中吸收营养。他们观察到，灯架虎耳草及另一种食虫植物的对照植株中，标记氮的水平显著上升，这与非食虫对照组形成鲜明对比，从而支持了灯架虎耳草物种具有食虫行为的假说。

发表在《Journal of Neuroscience》期刊上的一项研究发现，资深观鸟者的大脑结构与普通人之间存在一些差异，或许正是这些差异让他们能更好地识别陌生鸟类。这或许表明，观鸟活动能重塑人的大脑，其作用机制与学习一门语言或一种乐器类似。这类活动或许能增强参与者的认知储备，即提升大脑抵御衰老、适应损伤的能力。 在学习或练习某项技能时，大脑会强化并优化相关神经通路，这种能力被称为“神经可塑性”，是人类发展专业技能的基础。为探究观鸟活动是否同样能重塑大脑，加拿大约克大学的研究人员对 48 名业余观鸟者进行了大脑结构与功能分析。受试者年龄在 22-79 岁之间，两组在性别、年龄和受教育程度上均相当，其中有一半经筛查测试被判定为资深观鸟者，而另一半为新手。

北极多地气温本周多次超过 32 摄氏度。北极圈附近的 Selagoncy 气象站记录到 33.9 摄氏度的高温；另一个气象站 Olenek 的气温也达到了 32.7 摄氏度。气象学家 Alan Gerard 表示，最近几年的研究表明，由于海冰减少的反馈作用，北极升温速度远超地球其它地区，“这是一个危险的反馈循环：随着海冰融化，北极反射的阳光减少，海洋吸收的热量增加。反过来又导致海冰进一步减少。该地区升温速度比世界其它地区快四倍。”今年预计将出现强厄尔尼诺现象，气温上升的趋势短期内不太可能放缓。

Rust 语言团队宣布了新的 LLM 政策。 除了作者没有人被要求必须阅读 LLM 输出，除非其自愿选择：LLM 输出不允许出现在公开文档、PR 描述或 GitHub 评论中，除非明确标注；审核者如果不愿意，无需查看 LLM 生成的 PR。 没有人必须使用 LLM 为 rust-lang/rust 做贡献：政策必须先为人类编写，仅为机器提供摘要；LLM 审核不能替代人工审核或自审。 你可以生成仅供自己查看的 LLM 内容，无需披露，前提是不会将其发布到任何你期望我们阅读或审核的地方。 你可以用 LLM 回答问题、分析、提炼、完善、检查、提出建议和审核，但不能用来**创造**。

科学家拍摄到迄今为止​最清晰的太阳表面图像，展现出比以往所见复杂得多的太阳表层景象，为该层的磁活动区提供了新的见解。研究报告发表于《自然》期刊。太阳的可见表面又称光球，是其大气的一个薄层。等离子体在这里变得透明，使光能够逃逸。已知这一区域动态且复杂，磁场和对流等众多因素塑造了太阳的大气结构。美国国家太阳天文台（NSO）的研究人员利用位于夏威夷的井上太阳望远镜，获取了具有迄今最高空间分辨率的太阳可见表面观测结果。该研究观测了一处靠近太阳黑子的磁活跃区域。在新的高分辨率图像中，太阳表面主要呈现带状和涡旋状结构。研究人员认为，这些特征与开尔文-亥姆霍兹不稳定性有关，后者是一种经典流体动力学现象，当两种流体以不同速度互相掠过时就会形成波浪状结构。这是首次获得证据，表明太阳表面存在这一现象。

DigiTimes 报道，受 AI 需求和长期协议的推动，三大内存制造商三星、SK 海力士和美光 2027 年的 DRAM 和 HBM 内存产能已全部售罄。PC、笔记本电脑和智能手机使用的 DRAM 内存供应在 2027 年将更紧张，NAND 闪存的产能也可能在 2026 年 8 月底前被全部预订。这三大内存制造商占据了逾九成的市场份额，它们没有新增供应的计划。AI 对内存的需求远超过供应，内存制造商正与企业签订长期协议以确保长达 5 年的内存供应，情况相当严峻。

Google 对其 AI 部门 DeepMind 进行了领导层重组，诺贝尔奖得主、DeepMind CEO Demis Hassabis 卸任，他将担任新设立的首席科学家职位并改任董事长。资深工程师以及 Google Brain 联合创始人 Jeff Dean、Sanjay Ghemawat、Oriol Vinyals 和 ​Quoc Le 都离开 Google，他们成立了一家公益性 AI 企业 Discovery Loop，专注于机器学习、科学和工程领域的突破性研究。这次人事变动恰逢 Google 最新 Gemini 模型的发布滞后，原计划 6 月发布，但至今仍未发布，引发了 Google 落后于竞争对手 Anthropic 和 OpenAI 的担忧。

为了推广 AI 工具，许多企业将 AI 使用率作为绩效考核的一部分。结果就是员工为了绩效致力于最大化 AI 使用，导致企业很快发现 token 费用大幅超出预算。软件巨头微软成为最新一家建议工程师限制使用 AI 的公司。微软执行副总裁 Jay Parikh 在一封发给微软员工的邮件中要求工程师专注于业务成果，而非最大化 AI token 的使用量，为了“从 token 投资中获得更大的价值”，微软将比其它模型更便宜的 OpenAI GPT-5.6 设为内部使用的默认模型。自 2026 年 7 月起，微软各部门将设定“AI token 预算目标”，员工可以追踪各自的 AI 支出。

从 5 月下旬到 6 月初，近 16 万名考生参加了墨西哥最大大学 UNAM 的入学考试。这是 UNAM 首次采用完全远程考试。结果是一场灾难。考试成绩公布后，高分比例比往年多多了。UNAM 120 道题考试 2021-2025 年间，只有 3.5% 的考生获得 100 分或以上，今年这一比例高达 16.3%；只有 0.9% 的考生获得 110 分或以上，今年这一比例达到了 5.5%。高分激增引发了考试作弊的指控。UNAM 的专家委员会在调查之后认为最佳方案是线下复试。受影响的 58,000 人将需要在监考人员监督下参加复试。

Waymo 联席 CEO Dmitri Dolgov 在 Y Combinator 的 Startup School 发表演讲，解释为什么自动驾驶汽车需要的传感器不能仅限于摄像头。特斯拉汽车只配备了光学摄像头，它的辅助驾驶系统依赖于来自摄像头的数据。Dolgov 解释说，人类仅靠眼睛就能驾驶汽车，如果自动驾驶系统的目标是实现人类水平的驾驶，那么只靠摄像头可能够了，但上限也就是人类水平，而无人驾驶汽车被寄希望有更高的安全标准。相比下，Waymo 的自动驾驶汽车使用了摄像头、激光雷达和雷达三种传感技术。摄像头提供高分辨率和彩色图像，它们是被动传感，在黑暗和强光下性能会下降。激光雷达直接测量世界的三维结构。雷达能穿透雾、雨、雪，利用多普勒效应直接读取速度。激光雷达和雷达都是主动传感器，在漆黑的夜晚或刺眼夕阳下也能清晰探测物体。三种传感器并不是互为冗余，而是组合成一个完整的图像。摄像头如果粘了树叶，那么驾驶系统可能就会停止工作。Waymo 的三种传感器可以确保汽车在恶劣天气下回家。

Telegram 因有用户分享 CSAM（child ​sexual abuse material）材料而被苹果在全世界短暂下架。苹果发言人证实了此次短暂下架事件，表示苹果的审查发现该应用存在违反禁止 CSAM 材料的内容，“在开发商迅速删除相关内容并封禁发布该内容的用户后，该应用已恢复上架。”Telegram 有逾 10 亿用户，它表示对 CSAM 内容采取零容忍政策，今年已因此封禁了近 33.8 万个群组和频道。英国监管机构 Ofcom 今年四月因类似的原因对 Telegram 展开了调查。Telegram 则坚称它没有 CSAM 问题，称自 2018 年以来已通过检测算法几乎完全杜绝 CSAM 材料的公开传播。

中国创新药研发快速发展，带动实验猴需求激增、价格接近翻倍，而这造成的供应紧张可能拖慢新药试验进度。今年 6 月一家国家级实验室以每只17.8 万的价格采购 40 只食蟹猴，价格较一年前接近翻倍。下一代癌症疗法开发商 Excalipoint Therapeutics 联合创始人兼首席财务官朱杰伦预计，实验猴明年每只售价可能突破 20 万元，较一两年前的略高于 10 万元接近翻倍。部分药物在获准进入临床试验前，必须通过猴体试验评估安全性。灵长类动物与人类生理结构相近，可用于观察药物在体内的运行及对器官的影响。单个生物药研发项目可能需要十几只至 100 只实验猴。业内人士和分析师指出，实验猴养殖场供应总体稳定，价格上涨主要是因为生物药和下一代疗法大量涌现，导致涉及灵长类动物的试验需求远超现有承载能力。目前中国约占全球创新药研发管线的三分之一，并已成为全球临床试验的首要目的地。