OrangeBot.AI Digest — 2026-09-06

Know the moment a domain opportunity appears

Search, organize and export every AI chat in one place

Agentic video analysis for faster, smarter Gemini insights

Your coding agent runtime. Claude but fast, cheap, concise.

fal's post-trained MiniMax H3 for quality video production

Create beautiful docs, by typing. No AI required.

The AI employee in Slack that reads the room.

Open source AI gateway turning traffic into a better model

Behavioral learning that makes AI agents better over time

Review code with your coding agents before committing

Live commute times in your menu bar

Turn family voices into hand-drawn story films

Markdown feature flags your coding agent installs for you

Turn any prompt into a brick set you can actually build

Lets your AI agents debug production without redeploying

Many recurring text functions are easy to describe but difficult to implement with rules, while calling a large remote model for every input introduces repeated cost, latency, and dependency on a provider. We present compile by training, which turns a natural-language specification into a reusable neural function. At compile time, teacher models generate task-specific examples that are used to train a small adapter for a compact interpreter. The resulting function runs without the teachers and can be stored, versioned, and composed like ordinary software. On FuzzyBench-Hard, a subset on which the Program-as-Weights fast compiler produced no exact matches, compile by training reaches 83.6% semantic accuracy. This higher accuracy comes with a higher compile-time cost: roughly a minute rather than seconds for the fast compiler. We deploy the compiler in a public interactive service and demonstrate compiled functions in a multi-site website helper, a language-controlled 3D avatar, and a bidirectional English-Claudish translator.

As terminal-based code agents become prevalent, agent trajectories have accumulated at scale, while realistic, executable environments remain scarce. However, environments are what agent post-training actually requires: each can be re-queried into many verifiable tasks and provides execution feedback, whereas a trajectory is a single frozen demonstration. Rather than generating environments from scratch, we observe that the tool-execution history in existing trajectories exposes the structure and contents of the environments in which they ran, making it possible to reconstruct those environments from the trajectories themselves. Thus, we introduce Terminal-Universe, a framework which turns each trajectory into a reusable environment and explores it for synthesizing new tasks and continued interactions. Specifically, Terminal-Universe replays the file operations recorded in a trajectory to restore each file before the agent modified it, yielding a partial workspace; a completion agent then supplies the missing files and dependencies. On this recovered workspace, we both reconstruct the original intent task and synthesize entirely new ones. Besides, we also scale the tasks along two complementary axes: breadth and depth. For breadth, we mine directional dependency relations between related environments and synthesize cross-workspace queries spanning multiple codebases, as developers routinely do in real-world development. For depth, we extend the initial single-turn query into a multi-round session that captures iterative user feedback and requirement refinement via a user agent. Applied to public terminal agent trajectories, Terminal-Universe produces 37.3k task-sufficient environments. Supervised fine-tuning of Qwen3.5-27B on this corpus improves single-round performance on Terminal-Bench 2.1 by 11.9 points and multi-round performance on EvoCode-Bench v2 MT@4 by 13.8 points.

We introduce LLaDA-Image, a unified framework that pairs a 6B Diffusion Transformer (DiT) trained from scratch with a frozen vision-language understanding module built on the LLaDA2.0-Mini diffusion language model backbone. Instead of relying heavily on paired image-text data from the beginning, we first build a strong visual generative prior through image-only pre-training and mid-training. The generation pipeline comprises 220M samples, 98 of which are real images. For efficient and scalable optimization, we use parameter-free RMSNorm throughout the DiT together with the Muon optimizer. The resulting unified model produces highly photorealistic images while accurately following fine-grained editing instructions. We further distill LLaDA-Image into LLaDA-Image-Turbo, enabling fast inference in 2-4 sampling steps. On Qwen-Image-Bench, LLaDA-Image achieves overall scores of 53.53 and 53.38 on the English and Chinese tracks, respectively, setting a new state-of-the-art among open-source models on both tracks. To support further research on capable and efficient generative models, we release our model weights, training code, and detailed recipes.

Large language models achieve superior performance on tasks that require extended reasoning, but long chains of thought make the KV cache a severe memory bottleneck. Existing KV cache compression methods share one paradigm: score each cached token by some estimate of how much it will matter later, and keep the top-scoring ones. We show that the selection signal contributes almost nothing. Random Attention keeps the prompt and evicts uniformly at random within each attention head, computing no score at all; across four models and six reasoning tasks it matches the strongest prior evictor while serving 32-43% higher throughput than it in vLLM deployment. Controlled experiments explain this by showing that 1) the prompt is the fragile part of the cache, and most of the gap between selectors is just whether their selection signal happened to keep it; 2) the reasoning trace protects itself against eviction with redundancy at two levels, in the text (the model restates what it still needs as it works) and across attention heads (each keeps its own copy of the trace), so once the prompt is safe, a random draw retains enough copies of what the model still needs, and no score is required to pick them. Our code is publicly available at https://github.com/SalesforceAIResearch/Random-Attention.

Large language models offer broad capabilities, but adapting them to evolving domains, tools, and requirements often entails repeated post-training. Autonomous systems automate parts of this process by proposing updates, training candidates, and using evaluation feedback to select subsequent proposals. As evidence accumulates, a central problem emerges: which past update evidence remains actionable after subsequent training has changed the parent model? An update's effect depends on its parent, data, and training stage. Treating past success as context-free permission can waste compute. If the resulting child is promoted, it can also degrade the subsequent training trajectory. We formulate this problem as conditional experience transfer and introduce Boundary-Calibrated Intervention Transfer (BCIT), a method that authorizes experience reuse before weight-changing training. BCIT binds an observed effect to its source context, checks applicability conditions, vetoes candidates with named hard conflicts, and obtains current-state evidence through a bounded training trial when needed. Fully trained candidates still face a shared adoption rule, and only observed events extend memory. On one 4B model adapted across finance reasoning, text-to-SQL, and function calling, candidate updates exhibit heterogeneous target and retention effects across the evaluated contexts. Under matched candidates, evidence, and compute, BCIT authorizes fewer harmful updates and attains higher equal-budget final-model quality than the evaluated alternatives. These results support treating experience authorization as a distinct problem in autonomous post-training.

Robot learning increasingly depends on broad and diverse demonstrations, yet collecting robot data remains expensive and poorly suited to covering the long tail of real-world tasks. To address this bottleneck, we introduce RoboTok, an internet-scale data engine that, given a query human manipulation video, retrieves manipulation-relevant human demonstrations from web videos for training dexterous robot policies. Specifically, we learn a latent motion space from 3D hand trajectories expressed in estimated actor-centered reference frames. This representation enables manipulation behaviors to be compared across variations in camera viewpoint, scene appearance, and actor occlusions, while remaining compact enough for efficient search and continual indexing over internet-scale video collections. We evaluate RoboTok against existing robot-data retrieval approaches on retrieval benchmarks and downstream robot policy performance. Our results show that RoboTok retrieves more relevant manipulation demonstrations and improves downstream task success, establishing hand-pose trajectory-aware retrieval as a way to make web video a scalable and continuously growing source of supervision for robot learning.

Compressed context is usually carried as human-readable text or as rendered images that must be decoded, even when its consumer is a language model. We introduce LatentPress, which writes conversational histories and long documents into a third representation: continuous memory tokens that a frozen decoder reads directly through its input-embedding interface, with no text reconstruction at inference. A small reader-matched writer compresses 4-16times while training only an adapter (4.2M-26.2M parameters, sim!0.1% of the decoder). On LongMemEval, LatentPress reaches 0.504 accuracy at 7.70times compression versus 0.490 for uncompressed evidence, outperforming text summaries (0.184) and OCR-based compression (0.426 to 0.312). On LongBench-QA, in-domain writers match or exceed raw-context reading at 4-8times compression, while 16times trails raw. Writing takes 43ms per conversation, roughly an order of magnitude faster than text summarization or OCR reconstruction, and reading is 5-9times faster than raw context or cached OCR. We validate the interface under two transfer settings, zero-shot from UltraChat to LongMemEval memory QA and from LongMemEval-derived QA to unseen LongBench document domains, establishing direct soft tokens as a practical machine-facing context interface beyond text and vision. The implementation of the experiments could be found at: https://github.com/xuyd16ai/context_softtoken_compress .

On-policy distillation (OPD) combines student-generated rollouts with dense token-level supervision from a teacher. Existing work has mainly studied its algorithmic behavior, leaving the role of training data unclear. We examine this role at the data-minimal limit by training on a single query. One-shot OPD keeps improving for hundreds of steps and recovers most of full-data OPD's gain across task domains and model families. We explain this result through the states visited during training and the rate at which the student aligns with the teacher. We measure state coverage, the fraction of the states full-data OPD visits that a query set's rollouts reach. A single query already reaches \(71.5\%\), most of it within the first 100 steps. Adding semantically distinct queries raises coverage and validation accuracy together, until 16 queries reach \(98.9\%\) and match full-data training. Yet alignment slows at a similar pace whether OPD trains on one query or the whole dataset, and even a fixed set of states takes hundreds of steps to absorb. OPD is therefore data-overfed but algorithm-starved. Its rollouts quickly expose broad supervision, while the student absorbs that supervision increasingly slowly. The state-coverage result extends to multi-teacher OPD, where 16 semantically diverse queries per domain match full-data MOPD. As a further stress test, content-light templates and off-domain WildChat queries also approach the real-query baseline. Task content and induced state coverage can therefore come apart. We hope these findings direct future work toward the step efficiency of OPD, and prompt a re-examination of the data and the mechanisms behind its recent successes in frontier post-training.

Hybrid LLMs pair softmax attention with linear-attention layers such as Gated DeltaNet (GDN), whose recurrent state summarizes the context in fixed size. Early community 4-bit quantizations of Qwen3.8-27B (48 GDN layers, 16 attention layers) left the GDN block in 8- or 16-bit precision -- especially its decay and write-strength gates -- on the intuition that errors in a recurrence accumulate over long contexts. We test that intuition by building Minima: NVFP4 W4A4 on all 496 linear layers, GDN included. Across perplexity at 4K/32K, MMLU-Pro, GSM8K, AIME'25, GPQA-Diamond, LiveCodeBench, and RULER retrieval to 64K, Minima matches BF16 within seed noise (5-task average -0.52) while being the smallest (17.5 GiB) and fastest-prefill (+14-19%) recipe we compare, and its 32K perplexity gap shrinks with position. A four-part mechanism study explains why: (i) NVFP4's 16-element block scaling localizes the residual stream's extreme outliers, equalizing activation error across layer roles; (ii) the supposedly fragile gate projections are the least sensitive -- softplus/exponential and sigmoid parameterizations compress ~11% GEMM error to ~2% output error; (iii) the delta-rule recurrence holds injected noise at a flat plateau over 32K tokens and forgets a state impulse within hundreds of steps, because each write overwrites the state along the current key direction; (iv) the per-token quantization cost washes out with context instead of compounding. We also repair a global-scale mismatch that arises when per-module-calibrated NVFP4 checkpoints are served by kernels that fuse those modules into one GEMM, and show calibrated FP8 KV-cache scales are performance-free. The result: a practical recipe -- quantize everything, ship KV scales -- and a mechanistic account of why the recurrent half of a hybrid LLM is the easy half to quantize. Checkpoint: https://huggingface.co/minima-ai/mnma_qwen3.8_27b_nvfp4

We propose Puffin-World, a unified multimodal architecture that integrates physical understanding, spatial simulation, and 3D world generation and reconstruction without relying on external offline modules. To reliably construct and interact with 3D worlds, our framework jointly models three native world states: physics (gravity field and latitude), geometry (depth), and appearance (image), together with a unified Omni-Camera representation that supports diverse tasks and flexible motions. Beyond modeling these states, we introduce a strategy for propagating physical dynamics across future frames. By grounding absolute camera properties in the real world, Puffin-World enables physically consistent and visually stable world generation. We further couple appearance and geometry within a single generative process, jointly synthesizing each future view and reconstructing its underlying geometry. This unified paradigm enables interleaved closed-loop applications requiring synergy across multiple tasks, including mimic and self-calibrated world exploration. To scale Puffin-World to complex scenarios, we construct Puffin-16M, comprising 15 million vision-language-camera triplets and 1 million trajectories featuring various and challenging motions. To foster further research in this area, we released the code, models, and datasets.

Online 3D reconstruction models perform poorly on long videos. This happens because regressing poses relative to a fixed first-frame anchor forces extrapolation far beyond the training distribution. Small drifts accumulate and amplify into significant geometric collapse. However, we observe that per-frame depth remains stable throughout this failure. The backbone's local geometry remains intact; only the global pose head breaks down. Motivated by this decoupling, we introduce Scal3R. This approach reformulates online reconstruction as multi-reference relative pose querying. We use lightweight learnable tokens, which make up about ~1% of the parameters, and inject them into a completely frozen backbone via asymmetric attention. This setup queries poses relative to multiple past keyframes. An online pose-graph optimization system with loop closure suppresses long-range drift. Scal3R reaches convergence in 8 hours on a single GPU. It reduces the average ATE by over 60% on KITTI compared to the online baseline. It also achieves state-of-the-art performance across Virtual KITTI, Sintel, TUM-Dynamic, ScanNet, and 7-Scenes. Project page: https://linjohnss.github.io/scal3r/

While diffusion base models such as GPT-Image-2 and Nano-Banana exhibit remarkable visual expressiveness, their end-to-end generation inherently yields flattened bitmaps with error-prone text, precluding layer-wise post-editing. Conversely, code-based visual generation via Coding Agents provides precise layout control and decoupled layers, yet remains constrained by a lack of global aesthetic intuition and the difficulty of coding complex visual assets. To address this, we propose Editable Visual Design, a new paradigm driven by a Coding Agent. We designate the VLM as the ``creative brain'' for requirement comprehension, task planning, and aesthetic judgment, while utilizing the image generation model as an on-demand ``visual world simulator'' to synthesize standalone visual assets. Operating under an ``imagine first, then act'' closed-loop workflow, the agent generates isolated assets, writes native HTML/CSS, and iteratively refines the design against visual rendering feedback. Furthermore, Agent Design Replay faithfully reproduces the creative and reasoning trajectory akin to that of professional human designers. Ultimately, the system delivers editable artifacts with decoupled layers and real text, enabling users to perform intuitive mouse dragging and layout adjustments on a graphical user interface. Validations on posters, infographics, and other scenarios show that this paradigm successfully achieves both refined aesthetics and production-grade editability.

Joint audio-video generation models have made substantial progress in visual quality and audio-visual synchronization. However, they still provide limited control over when shot transitions occur and dialogue is spoken. This limitation constrains their application in script-driven content creation, where timing errors can undermine narrative coherence and the viewing experience. Current joint generators align video and audio representations on a shared temporal axis, yet the precise timing of shots and dialogue specified in a structured prompt is encoded only in the prompt's text representation and remains unaligned with the temporal coordinates of either modality. Consequently, video and audio may remain synchronized with each other while both fail to follow the script timeline. This mismatch motivates us to extend temporal alignment beyond video and audio to include the structured script. We therefore introduce Temporal Context Routing (TCR), which maps the script timing onto the shared temporal axis of video and audio generation and routes each prompt's guidance to the corresponding positions in both modalities. Compared with the baseline on 200 test scripts, TCR reduces Shot Boundary MAE by 96%, from 1.11 s to 0.042 s, and raises Dialogue [email protected] s from 28.3% to 84.1%. TCR achieves these improvements while maintaining visual quality and audio-visual synchronization comparable to those of the baselines. A user study further shows that participants prefer TCR on all five evaluated dimensions.

For scientific progress, we need benchmarks that test the limits of state-of-the-art models, and evaluation methods that inform us about failure cases. As models get stronger, standard benchmarks for machine translation are approaching saturation. Further, automatic translation metrics are unreliable, vulnerable to reward-hacking, and provide unactionable assessments. Even gold human evaluation is not problem-free, because it often lacks reproducibility, objectivity, and scalability. Overall, this prevents us from tracking objective progress in the field and identifying pathways for improvement. We introduce the Last Translation Benchmark, a collection of human-authored and peer-reviewed examples (texts, images, audio, videos) that break leading machine translation models. We also present a new evaluation approach: each example comes with handcrafted verification rules describing concrete failure cases on that example, therefore allowing reliable and actionable future evaluation. The Last Translation Benchmark is a live dataset that accepts ongoing contributions. The latest version is LTBv1, containing accepted contributions prior to September 1st 2026, with future releases planned as new data is continuously collected.

Streaming video understanding requires multimodal large language models (MLLMs) to process continuous visual inputs and respond to user queries under strict causality and bounded memory. Existing approaches typically compress historical observations into an external memory bank and retrieve query-relevant evidence as additional visual context. Though effective, this store-and-retrieve paradigm keeps historical evidence as external visual context, preventing it from being internalized into a compact, evolving latent memory that can continuously guide streaming reasoning. To bridge this gap, we introduce LatentStream, a progressive latent working memory framework that shifts streaming memory from store-and-retrieve to retrieve-and-internalize. Specifically, LatentStream comprises three coordinated components. First, Query-agnostic Hierarchical Streaming Memory organizes visual history into short-, mid-, and long-term levels under a fixed memory budget through Jenks-guided adaptive consolidation. Once a query arrives, Hierarchical Latent Memory Evolution equips groups of latent memory tokens with progressively expanding memory receptive fields, enabling them to iteratively retrieve historical evidence from their corresponding scopes and internalize it into a compact, fixed-length latent memory. Finally, Progressive Confidence-guided Latent Memory Optimization constructs a hierarchical progression reward from group-wise predictive entropy and jointly refines the latent memory tokens and retrieved evidence, encouraging increasingly confident streaming reasoning. Extensive experiments demonstrate that LatentStream achieves new state-of-the-art results on existing online and offline video benchmarks.

OpenAI : OpenAI Chief Scientist Jakub Pachocki says no lab has solved alignment enough to keep scaling at maximum speed, and hopes voluntary slowdowns become commonplace —  Author: Jakub Pachocki, Chief Scientist at OpenAI  —  In mid-2023, within the “RLSlow” research project …

Valida Pau / The Information : Analysis: since October, Anthropic has entered into agreements for at least 14.8 GW of compute capacity and may spend as much as $517B over the next decade —  Anthropic in the last year has scrambled to line up cloud computing deals with SpaceX, Google and others to meet the skyrocketing demand …

OpenAI : OpenAI says it hit its “automated research intern” goal, its researchers now use 3.1 agent-workdays per human workday, and top users spend $7,000+/day on tokens —  For AGI to benefit all of humanity, we believe it must be democratically governed.

Mark Gurman / Bloomberg : Tim Cook's pay package indicates he will remain very active at Apple; sources: Apple's plans to squeeze more App Store profits prompted Phil Schiller to leave —  Also: The latest on Apple's big Sept. 9 event.  —  The early signs out of Cupertino are that Tim Cook will still loom large at Apple.

New York Times : How AI gutted Kenya's essay-writing industry, which at its peak paid 40,000+ people in Nairobi to do overseas students' homework, leaving few paths back to work —  Thousands of Kenyans made a living writing essays for overseas students.  With A.I., the work has dried up, a warning for online gig work that has been a global lifeline.

Philip Wegmann / Wall Street Journal : Trump officials say Judeo-Christian principles inform AI policy, with some optimistic about AI and others seeing parallels to biblical end-times prophecies —  Trump administration officials are wrestling with the spiritual consequences of AI as the president embraces it as an economic engine

Rafe Rosner-Uddin / Financial Times : Sources: Travis Kalanick's Atoms is developing robotaxi tech and hired Anthony Levandowski after acquiring his company Pronto; Uber has invested $100M in Atoms —  Atoms gears up for hiring spree and could link with Uber, the company he was ousted from almost a decade ago

Emily Forlini / Fortune : OpenAI quietly updates its evaluation metrics for GPT-6 Astra, making changes that appear to favor Astra and continuing to revise other metrics after launch —  OpenAI has changed several evaluation benchmarks for its GPT-6 Astra model since first publishing a blog post announcement mid-afternoon on Sept. 3.

Stephanie Findlay / Financial Times : The data center backlash is challenging Texas' pro-business approach; Wood Mackenzie: Texas has more data center capacity under construction than any US state —  For 30 years the state has profited from going the extra mile for business.  Data centres are putting that to the test.

Morgan Meaker / Bloomberg : As researchers begin applying AI to understand animal communication, bioethicists warn it could give humans new ways to manipulate, exploit, and harm animals —  AI could transform our understanding of other species.  It also risks giving humans powerful new ways to manipulate, exploit and harm them.

Todd Bishop / GeekWire : The Seattle Times and Newsday sue OpenAI and Microsoft, alleging the companies trained AI on their journalism; Microsoft and OpenAI are funders of Seattle Times —  Microsoft was sued Friday by the parent company of its hometown daily newspaper, The Seattle Times Co., which joined with Newsday …

ABC News : Trump admin launches arcade.gov with a deportation-themed Snake game clone and a Tetris-like game; Tetris says it takes “copyright infringement very seriously” —  In “Build the Wall,” players aim to “protect the border” from zombies.  —  Isabella Murray, Nicholas Kerr, and Michelle Stoddart

Jean Eaglesham / Wall Street Journal : Swiss Re: global premiums for insuring data centers will likely reach $20B-$30B/year by 2030; ~40% of US data-center capacity is located in tornado-prone areas —  AI hyperscale data centers are exposed to a dizzying array of potential threats during construction and operation

Miranda Murray / Reuters : Berlin is reviewing Rhysida's 5.79TB release of state data after refusing to pay a ransom; files reportedly include national defense and threat response plans —  Berlin's state government said on Saturday it was reviewing with the highest intensity a trove of stolen data published by a ransomware group …

Robert Hart / The Verge : Anthropomorphic portrayals of AI models as rogue agents can obscure the responsibility that companies like OpenAI have for incidents like the Hugging Face hack —  The internet fights over anthropomorphism around the Hugging Face hack. … Depending on who you ask, developer platform Hugging Face …

根据发表在《Ambio》期刊上的一项研究，观鸟有益身心健康。Project FeederWatch 项目的参与者中有 95% 表示喂鸟和观鸟对其幸福感产生了积极影响。相比之下只有 55% 的参与者表示同期生活中其他事件对其幸福感产生了积极影响。参与者表示，喂鸟时会感到快乐、感激、充满希望和内心平静。一部分人表示观鸟有助于减轻压力，或者在生病或家人去世等困难时期将注意力集中在积极的事情上。还有人表示观察喂鸟器激发了好奇心，有助于保持专注，或者帮助培养鸟类识别技能。

《美国膳食指南》建议每周至少食用 8 盎司鱼，推荐人们选择三文鱼（salmon）等富含 Omega-3 脂肪酸的鱼类，Omega-3 脂肪酸被认为有益心脏健康。这一推荐导致三文鱼产量大增，而人类消费的近七成三文鱼如今都来自养殖渔场。根据一项新研究，养殖三文鱼的 Omega-3 脂肪酸含量大幅下降，意味着相关推荐数据已经过时。Omega-3 脂肪酸含量下降的原因是三文鱼的食物结构发生巨大转变，从肉食转变为以大豆和菜籽油为主食。世界最大的养殖三文鱼生产国挪威，1990 年九成的饲料原料来自海洋，但到 2020 年这一比例降至约 23%。挪威养殖三文鱼饲料中九成的大豆来自三家巴西公司。

第 84 届世界科幻大会上周公布了 2026 年雨果奖获奖名单，25 岁的美国作家 H. H. Pak 超过乔治·R·R·马丁（George R. R. Martin）成为最年轻的最佳中短篇小说奖得主，马丁是在 26 岁通过中短篇小说《A Song for Lya》赢得雨果奖。获奖名单包括： 最佳长篇小说：Alix E. Harrow 的《The Everlasting》，Netflix 已宣布将其改编成电视剧； 最佳中长篇小说：Amal El-Mohtar 的《The River Has Roots》； 最佳中短篇小说：H.H. Pak 的《Never Eaten Vegetables》； 最佳短篇小说：Thomas Ha 的《In My Country》： 最佳系列小说：John Scalzi 的《Old Man’s War》系列； 最佳科幻电视剧：Apple TV+ 的《Murderbot》第九集《All Systems Red》； 最佳电影：《Sinners》： 最佳游戏：《Clair Obscur: Expedition 33）》。

科学家利用欧洲 450,112 名参与者（131,426 名男性 和 318,686 名女性）的数据，评估了红肉、加工肉和白肉与胃腺癌和食管腺癌之间的关联。研究人员对红肉、加工肉和白肉的定义是：红肉包括猪肉、牛肉、小牛肉、羊肉等；加工肉包括火腿、培根、香肠、加工肉块、汉堡包、肉丸、肉酱等；白肉包括鸡肉、鸭肉、鹅肉、兔肉以及其它禽肉。结果显示，每天每摄入 30 克加工肉，胃癌风险增加 9%，食管癌风险增加 13%；每天每摄入 20 克白肉，非贲门胃癌风险增加 12%。

物理学有两大不兼容的理论，其一是爱因斯坦广义相对论，适用于宏观物理领域；其二是量子力学，适用于微观物理领域。爱因斯坦曾尝试将两大理论融合成一个统一的理论，但没有成功。广义相对论的一个核心概念是等效原理，它认为引力和加速度不可区分。等效原理指出，对于自由落体的观察者，引力在局部应该消失。例如在电梯中自由落体的人会体验到失重。根据发表在《Science Advances》上的一项研究，科学家利用冷却到绝对零度附近的铷原子云，将其置于量子叠加态，通过以伽利略名字命名的量子干涉仪，在实验中测试了量子领域的等效原理。结果显示等效原理适用于量子领域，观察结果与引力如何影响量子波的预测相吻合。研究本身并未统一广义相对论和量子力学，但它朝着该目标迈出了一步。

两年前，蒂姆·安德鲁斯疲惫不堪，唯一想做的就是睡觉。糖尿病和高血压毁掉了他的肾脏，而通过过滤血液维持他生命的频繁透析治疗令人精疲力竭。尽管只有 66 岁，他却毫无精力与食欲，虚弱得无法行走，还曾两次心脏病发作。肾移植本可以救他的命，但器官供应短缺。根据美国器官共享联合网络的数据，约有 9 万人在等待肾移植，平均每天有 11 人在等待中去世。而且由于安德鲁斯的血型较为罕见，他获得匹配的人类肾脏的几率尤其渺茫。当他听说麻省总医院布里格姆的医生正在为患者移植转基因猪的肾脏时，他主动联系了他们。“我想，如果我要死了，不如为人类做点事，参与这个实验，”他说。2025 年 1 月 25 日，安德鲁斯接受了转基因猪肾的移植，带着它生活了九个月。今年他成为已知首例在猪肾移植后接受人类供体肾脏的患者，周四发表于《柳叶刀》的一篇论文中描述了这一里程碑式的成就。他的病例为移植医学中一个引人关注的概念提供了初步证据：猪肾虽尚非永久性解决方案，但可以作为通往人类器官移植的“桥梁”。这些猪肾来自经过大量基因编辑的猪，其中一些编辑旨在防止严重的器官排斥反应，还有一些用于灭活猪基因组中的病毒。

一周前，Debian 项目经过投票允许以负责任的方式使用生成式 AI，Debian 项目表示，它既不反对也不支持在软件、包、文档等的开发和维护中使用生成式 AI 工具。但项目也认识到，如果能负责任的使用 AI 工具，将能显著提高贡献者的效率，使他们将有限的时间投入到需要技术专长、判断力、审核和协作的工作中。现在，Android 自由软件应用商店 F-Droid 考虑采用与 Debian 相同的 AI 政策，它的政策提议直接拷贝了 Debian 的政策全文，只是将 Debian 的名字替换为 F-Droid。

联合国会员国周五以 164 票赞成、1 票反对的结果通过了一项决议，倡导采用能更准确反映各大洲真实大小的地图投影法——这是由非洲主导、反对 16 世纪墨卡托投影法的运动所取得的成果，而墨卡托投影法至今仍是世界上使用最广泛的地图投影法。美国投下了唯一一张反对票，将该倡议斥为“激进意识形态项目”的一部分。爱沙尼亚、格鲁吉亚、立陶宛、摩尔多瓦、塞尔维亚和乌克兰等六个国家投了弃权票。该决议并未禁止使用墨卡托投影法，也未强制要求采用替代方案。相反，它鼓励各国政府、学校、国际组织和科技公司在相对大小至关重要时，使用“平等地球”投影法及其他所谓的等面积地图，并向公众阐明任何平面地图在表现球形地球时的局限性。法国已经宣布计划在其世界地图中放弃使用墨卡托投影法。

硅谷亿万富翁、Palantir 联合创始人 Peter Thiel 前不久携家人移居了阿根廷，并斥资 1200 万美元购买了一栋豪宅。许多人猜测他可能是为了躲避可能的三次世界大战，阿根廷地处南半球深处，农业资源丰富，是末日避难的理想之地。然而 Peter Thiel 在阿根廷深入参与了当地的政治活动。他与阿根廷总统 Javier Milei 及其亲密顾问进行了多次闭门会谈，之后 Milei 的政府提出了一系列符合 Peter Thiel 利益的提案，其中一项提案要求政府民事部门与国家情报机构共享阿根廷公民个人数据。很多人怀疑 Thiel 的 Palantir 公司可能与阿政府签订了未公开的合同。阿根廷众议院为此邀请 Thiel 和 Milei 前来回答一系列相关问题，但无人出席会议。随后阿根廷人在 Thiel 的豪宅前举行了抗议活动。

巴西科学家分析了 53 条当地捕获的金枪鱼，发现几乎所有鱼都携带了寄生虫。研究团队共统计到 1600 多条寄生虫；肾脏和心脏是唯一没有寄生虫的器官；除此之外寄生虫遍布全身；53 条鱼有 96% 携带了寄生虫；肠道是感染最严重的器官，其次是肌肉组织（即食用部分），然后是胃。巴西研究人员分析的是学名为 Katsuwonus pelamis 的金枪鱼品种鲣鱼，它是巴西最丰富的金枪鱼品种，也是 2022 年全球捕捞量第三大的海洋鱼类——当年全球共捕捞并销售了 310 万吨鲣鱼。

犹他州成为美国第一个要求 VPN 软件验证用户年龄的州。犹他州的年龄验证法律《Online Age Verification Amendments》明确将 VPN 软件纳入监管范围。这项法律规定，任何商业网站，如果逾三分之一内容被认为对未成年人有害，必须实施合理的年龄验证方法才能允许犹他州用户访问。只要用户的实际位置在犹他州内，不管是否使用 VPN 之类的方法隐藏地理位置，都需要验证年龄。违规网站将面临每次违规最高 2,500 美元的罚款。网站也被禁止发布如何使用 VPN 规避年龄验证要求的说明。EFF 认为这项法律可能迫使网站屏蔽所有已知的 VPN 地址流量，还可能违反了美国宪法第一修正案。

今年五月被捕的 Google 安全工程师 Michele Spagnuolo 被控利用内部信息在预测市场 Polymarket 押注歌手 d4vd 成为 2025 年 Google 搜索量最高的人物而获利 120 万美元。他被控犯有商品欺诈、电信欺诈和洗钱罪。 他的律师团队本周递交了一份动议，寻求撤销对他的指控。Spagnuolo 没有直接否认利用内部信息获利，而是宣称他只是在赌博，而赌博行为不是受美国《商品交易法》监管的金融工具，且赌博发生在美国境外，不受美国司法管辖。Spagnuolo 已被 Google 停职。Polymarket 虽然是一家美国公司，但其预测市场在美国被禁，其平台也是由名为 Adventure One QSS 的巴拿马公司运营。Spagnuolo 在 Polymarket 上押注时也不是在美国，而是住在瑞士苏黎世。Spagnuolo 律师团队声称他利用的内部信息对 Google 没有任何价值。

日本研究团队发布研究结果称，在往返于北海道和东亚的候鸟粪便中发现了耐药菌。这些细菌已被报告对多种抗菌药具有耐药性，可能正在跨国境扩散。团队 2024 年 9 月在北海道厚岸町的灰背鸥繁殖地采集了粪便进行分析。这种鸟夏季在北日本地区繁殖，一天平均在栖息地周边飞 25 公里；冬季则移动至 4000 多公里以外的东亚地区。分析结果显示，检测出了“嗜麦芽窄食单胞菌”。该菌会在免疫力低下者中引发肺炎和血流感染症，也被认为是导致新冠病毒感染重症化的原因之一。检测出的部分细菌对两种用于治疗该菌感染患者的药物表现出耐药性，通过蛋白质等的膜抵抗药物攻击的守护能力以及活动能力较高。分析遗传信息后发现，其中包含日本国内尚无报告案例的类型，有可能是从海外带入的。灰背鸥是杂食性鸟类。或许是它们在各处吃食的过程中，也把细菌摄入了体内。

虽然维基基金会拒绝自愿认可员工工会，但根据美国劳资关系委员会于 9 月 3 日公布的不记名投票结果：维基媒体基金会美国员工以 158 票赞成、14 票反对压倒性多数通过了组建工会的决议，有效票数占 91.9%，投票率 81%。维基媒体基金会发表声明，表示尊重结果，接下来将与工会展开集体谈判。

联合国大会周五投票决定是否淘汰传统的墨卡托投影法地图，改用一种能更准确展现非洲面积的地图。墨卡托投影法地图由比利时佛兰德地理学家杰拉杜斯·墨卡托在 1569 年绘制，一直是全球标准地图，其缺点是高纬度地区面积变形失真，纬度越高变形越严重，接近南北极的高纬度地区如格陵兰面积被失真放大数倍之多。地图显示非洲的面积与格陵兰岛相仿，但实际上非洲大陆的面积是其 14 倍。历史学家和地理学家多年来一直指出墨卡托投影法源于西方刻板印象，部分活动人士甚至创造了“制图殖民主义”一词。改变地图的决议由多哥提出，得到了非盟的支持。联合国决议不具有强制执行力，如果投票通过，可能会促使课程和日常技术更新地图。