OrangeBot.AI Digest — 2026-10-15

Cut AI agent costs and control runaway spending

Isolated cloud machines for your AI coding agents

Rediscover the days behind your favorite songs

Build AI coworkers for any workflow, any industry

Turn a plain English idea into a circuit board you can order

The AI Runtime Backend QA Engineer

GNOME 51, Rust coreutils, and a refreshed Linux desktop

A open-source human approval layer for AI Agents

Let ChatGPT & Claude answer customer support tickets via MCP

Catch risky pull requests before they break production

Think ahead. Fight less. A weekly family briefing for dads.

Z.ai’s legal AI model for contract review and compliance

Email that feels like chat

Agent memory with Laya, Jev + Google Gemma

Forward iPhone texts to Slack, Discord, or Telegram

Context can improve large language model performance, but these gains may be lost after context removal. Inspired by biological systems consolidation, we formulate Memory Consolidation (MC) as internalizing these gains while preserving non-target capabilities, connecting context-specified editing, skill learning, and behavioral steering under a common objective. We introduce a five-task-family testbed measuring consolidation and retention under single, batch, and sequential updates. We propose Cortex, which generates prospective queries to anticipate uses of context and distills context-conditioned outputs into mergeable low-rank or structural feed-forward network updates. Across 4B--120B dense and mixture-of-experts backbones, including both text-only and multimodal models, Cortex lies on the empirical consolidation--retention Pareto frontier in most evaluated settings. After 50 sequential factual edits on a 14B model, Cortex-Structural recovers 88.9% of the context-induced gain while retaining 99.8% of original non-target performance.

Large language models (LLMs) increasingly tackle complex real-world tasks in specialized domains. However, identifying and correctly applying domain-specific rules remains challenging when scenario information is distributed across multiple files. Existing rule-reasoning benchmarks typically represent scenarios using only a few factual statements, simplifying the identification and application of relevant rules. To address this gap, this paper introduces RuleTrek, a multi-domain benchmark for agentic rule reasoning in complex scenarios. Its rule-grounded construction pipeline extracts structured rule graphs, generates and incrementally composes rule-specific sub-scenarios while ensuring consistency and solvability. To systematically evaluate agentic rule reasoning in complex scenarios, RuleTrek assesses three capabilities in interactive workspaces: Rule Identification (RI), Rule Deduction (RD), and End-to-End Rule Reasoning (E2E). Experiments on 11 frontier LLMs show that the highest RI and RD scores are only 62.47\% and 60.43\%, respectively, while E2E scores range from 26.97\% to 55.29\%. These results demonstrate that current LLM agents face substantial challenges in rule reasoning over complex scenarios. We make our code and dataset publicly available at https://github.com/yuan247/RuleTrek{RuleTrek}.

A recommendation request on Reddit draws several replies, each naming different titles. When an LLM recommender is trained on such threads with reinforcement learning (RL), which replies should a generated list be rewarded against? The state-of-the-art recipe, Rank-GRPO, picks one reply per training example, so the advantage mostly records which reply was drawn (85% of its variance) rather than how good the list is. We argue that RL should train on the request, not on each answer, and propose ReMaRk: it scores every list once against the union of all replies' titles and samples each request equally. This makes RL cheaper. On seven backbones and three datasets, ReMaRk reaches the validation recall of 2,000 per-reply RL steps in 25 to 53% of the steps and leads at that budget; trained until neither improves, it is never significantly behind Rank-GRPO and in four of seven pairs reaches Rank-GRPO's best with fewer GPU-hours. The same reward helps on ProtoQA, where about 100 people answer each question, which points to a simple rule for RL with several valid human answers: reward them jointly.

Language models are becoming infrastructure, and infrastructure built in the open, by public institutions, on documented data, is the alternative to dependence on a few providers; Apertus shows this is viable at scale. Apertus 1.5 extends the fully open Apertus family from text to multimodal reasoning. Continued pretraining of the Apertus 1.0 models, on 4T additional tokens for the 8B and 2T for the 70B, adds image and audio understanding and extends the context window fourfold to 262,144 tokens; multimodality enters through the tokenizer, the language-model head remains restricted to text, and the architecture is unchanged. Supervised finetuning, long-context finetuning, reinforcement learning with verifiable rewards, and preference alignment complete the released models. Thinking is a flag of the chat template rather than a separate model, and lifts mathematical reasoning by up to 20 points on the 8B model. The released models improve over their predecessors on nearly every benchmark; among fully open models of its class, the 8B leads on knowledge, instruction following, and cultural benchmarks, and its instruction following surpasses all compared models, including open-weight ones. On image and audio understanding, a first for the family, the models are competitive with fully open multimodal models, strongest on document understanding and long-form and lower-resource speech recognition, while specialized vision and audio models remain ahead on perception and transcription of English and other high-resource languages. As in Apertus 1.0, we train only on openly available data, respect crawler opt-outs retroactively, and exclude restrictively licensed material. We release the model weights, data preparation and training recipes, and evaluation suites under permissive licenses; openness extends to values, with the Apertus charter guiding alignment stages published alongside the models.

Vision-language models (VLMs) are increasingly deployed in multi-agent systems to automate complex computer-use workflows, yet existing benchmarks mostly assume consistent, cooperative environments. In contrast, real-world collaboration rarely runs on consistent information: peer or sub-agents frequently report outdated states, issue contradictory instructions, or compete over shared permissions and files, forcing agents to check others' outputs and verify ground truth in the environment. Evaluating such conflict resolution remains challenging: manual authoring cannot scale across complex systems, while unconstrained LLM generation produces narrative drift, answer leakage, and unsatisfiable rubrics. We introduce CLASHarness (Conflict-Loaded Agent Scenario Harness), a controllable framework for generating verifiable multi-agent conflicts in computer use, with a benchmark of 600 multi-round tasks. By combining hierarchical scenario planning with an eight-step verified micro-pipeline, CLASHarness pre-commits decisive visual facts before rendering and compiles hybrid code and semantic graders directly from executed reference trajectories, guaranteeing every task is solvable by construction. Evaluating 15 frontier models on CLASHarness reveals three core bottlenecks. First, when sub-agent outputs conflict with the environment, weaker models blindly follow misleading information, whereas stronger models inspect visual evidence to verify ground truth. Second, resolving conflicts is much harder than finding them: while agents can spot inconsistencies, actively fixing system state drops task success significantly. Third, coordination protocols remain a universal bottleneck: even frontier models fail over 58% of closing actions, often forgetting to release shared locks or hand off tasks. CLASHarness provides a scalable, verifiable foundation for building reliable multi-agent systems.

Hybrid reasoning models such as the Qwen3 series switch between a THINK mode and a NO-THINK mode through a few chat-template tokens. A popular recipe trains such models with reinforcement learning in NO-THINK mode and reports improved reasoning performance. We provide evidence that much of this gain can be explained by thinking leakage: the gain comes largely from partially reactivating the reasoning abilities of the original model's THINK mode instead of learning enhanced reasoning behavior. We train Qwen3-1.7B, 4B, and 8B with GRPO in NO-THINK mode. As accuracy improves, responses become longer and more reflective, are increasingly judged by the original model to resemble its own thinking, and the model's internal activations shift toward its THINK mode. We then investigate at the level of weights, activations, and attention heads: scaling the RL update further elicits THINK-mode behavior and approaches THINK-mode accuracy but does not exceed it; in Qwen3-8B, replacing a low-dimensional THINK-related activation subspace of the RL-trained model with its original-model value removes most of the RL gain, and the reverse exchange transfers a considerable fraction of the gain to the original model; and the RL-induced shift passes through the attention heads that read the empty think block. We further identify a possible mechanism in the GRPO training signal: responses favored by the advantage-weighted update are more THINK-like, and a gradient step computed from this signal moves the model toward its THINK state. Thus, gains in NO-THINK mode can largely reflect reactivation of reasoning abilities already present in the original model.

In an image latent space, the embeddings of high-resolution, natural, and sharp images form a manifold. Degradation of high-resolution images pushes their embeddings off this manifold. Real-world super-resolution (SR) then becomes the task of mapping the degraded embedding back onto this manifold --- not anywhere on the manifold, but to the point that preserves what the input still carries, both its semantics and pixel details. Every published method implements this mapping in a reconstruction-oriented latent space or pixel space. We claim these spaces are the wrong substrates for SR. Low-resolution and degraded images are embedded far from the manifold, making the mapping difficult and expensive. The lack of semantic information in these substrates also makes it difficult to navigate to the faithful point on the manifold, causing severe hallucination when degradation is heavy. Thus, restoring in a suitable latent space is crucial to the SR task. We show that the latent space of 23 fused layers of a frozen DINOv3-L is one such space that makes the SR task easier. Degraded images are embedded near the manifold. Moreover, this substrate contains a hierarchy of information, from pixel record to degradation robust semantics, guiding the model to find the faithful point on the manifold. On this substrate, a 415M decoder is trained under reconstruction and adversarial objectives to map the degraded embeddings back and decode to pixel space in one pass. The resulting model, RAESR, attains the best fidelity--perception trade-off among state-of-the-art adversarial and diffusion-based restorers on RealSR, DRealSR, LSDIR and DIV2K-Val, at 37 ms per 512 by 512 image on a single H20 GPU. Swapping the substrate for a VAE latent under an identical recipe loses on every metric.

Self-driving laboratories (SDLs) aim to automate the full experimental loop, from scientific decision-making to physical execution. Ideally, AI-generated plans could be carried out directly by robotic systems, but reliable automation remains difficult in complex, open-world laboratory environments, where experiments often involve fine-grained manipulation, long-horizon procedures, and substantial variation across tasks and setups. Humans therefore remain an important execution interface between AI-generated plans and physical experiments, creating a need for clear and effective human-facing experimental guidance. Because laboratory procedures are inherently visual, spatial, and dynamic, video is particularly well suited to communicating apparatus configurations, manipulation actions, temporal dependencies, and state changes. Recent advances in video generation now make it possible to synthesize experimental demonstrations directly from an initial workspace image and a natural-language instruction. However, whether such models can reliably communicate real laboratory procedures has not been systematically studied. We introduce LabInstruct, a benchmark for situated instructional video generation in real laboratories. LabInstruct contains 204 tasks across 5 scientific disciplines, with real reference executions and structured annotations of objects, actions, contacts, and state transitions. Evaluating 8 frontier image-to-video models, we find that visually plausible generations frequently remain procedurally incorrect, revealing a substantial gap between visual realism and the reliability required for experimental instruction.

Recursive self-improvement (RSI) in scientific research aims to improve agents' research capabilities through their own experimental experience. Scientific agent swarms can combine evidence and implementations from different research directions to guide their agents' subsequent experiments. However, poor experimental results can leave unclear whether to revise a hypothesis or its implementation, while overall program scores can hide components useful to other agents. To address these challenges, we introduce SwarmRSI, a framework that evolves scientific agent swarms for recursive self-improvement. We devise Feedback-Guided Swarm Mutation to refine hypotheses and their implementations using experimental feedback from other agents. We also propose Complementary Swarm Recombination to reuse retained hypotheses and complementary program components across agents, allowing partial results to support new research. Recursive Swarm Evolution integrates both mechanisms into a feedback loop, using new experimental evidence to guide subsequent research. Experiments on OpenProblems, NatureBench and MLE-Bench Lite demonstrate SwarmRSI's state-of-the-art (SOTA) performance. Further evaluations show that accumulated research experience improves performance on held-out tasks. Biologists also reproduce its programs for follow-up biological analyses.

In long-horizon tool-calling tasks, LLM agents can fail not only because of tool-use errors, but also because of ineffective interaction with users. Yet existing benchmarks largely assume cooperative and behaviorally stable users, leaving the effect of emotional variation on agent reliability underexplored. We introduce τ-Emotion-Bench, a ten-domain benchmark for evaluating tool-calling agents under controlled emotion-conditioned user interactions while preserving the underlying task objective. To scale both benchmark construction and agent training, we introduce Tracer, a scalable agentic data-generation pipeline that constructs grounded, policy-valid tasks and collects trajectories through an interactive agent--user--environment harness. By allowing dissatisfied users to terminate inefficient interactions, the harness provides an interaction-aware signal that filters trajectories with excessive user dependence and favors effective tool use. Experiments with two user simulators show that frontier agents remain unreliable, with substantial gaps between average task success and consistent task completion. Tracer-generated supervision consistently improves Qwen3 models across scales, generalizes to unseen domains, and transfers to τ-Bench and τ-Trait. Behavioral analyses further show that higher task success is associated with a better balance between tool execution and necessary user follow-up. Overall, τ-Emotion-Bench and Tracer provide a scalable framework for evaluating agent reliability under diverse user behaviors and improving tool-calling agents through interaction-aware supervision.

Professional agents increasingly work through command-line interfaces and multiple tools over shared workspaces containing files, messages, and business applications. These environments create safety risks that are difficult to capture with attacks placed at a single, known read location. We introduce APEX-Harm, a benchmark for evaluating prompt and script injection against long-horizon agents performing finance, consulting, and legal tasks in large workspaces. Each task contains 115 to 328 files, together with email, calendar, chat, document, spreadsheet, PDF, and code-execution tools exposed through MCP servers. The benchmark includes six attack categories covering static and dynamic injections, task-specific script replacement, and attacker-provided generic scripts, and evaluates attack success, exposure, and task performance. We evaluate 12 frontier models on 180 task instances per model. Opus 5 and Sonnet 5 have the lowest overall attack success rates (8% and 20%), while task-specific script replacement and attacker-provided generic script attacks remain substantially more effective, averaging 81-84% and 73% attack success, respectively. Gemini 3.8 Flash achieves the highest mean task score (0.877) while still exhibiting a 33% attack success rate. These results show that completing the requested task is not sufficient evidence of safe behavior: an agent can produce a correct deliverable while also performing an unauthorized action in the workspace.

Large language models (LLMs) are increasingly expected to orchestrate diverse tools over long-horizon tasks. Existing agents typically use Direct Tool Calling (DTC), where the model invokes tools sequentially, and tool responses accumulate in the model context, which can lead to context rot and orchestration breakdown. In contrast, Programmatic Tool Calling (PTC) invokes tools through executable code and stores intermediate results in runtime variables rather than appending them to the context. Prior work has explored PTC mainly as an inference-time optimization, but its benefits for agentic training remain unclear, and suitable verifiable training tasks are scarce. In this work, we introduce ProgAgent, a full PTC training stack, comprising a pipeline for synthesizing verifiable long-horizon tasks, a lightweight PTC runtime, and a two-stage training recipe. Using only 215 tasks instantiated at four environment scales, ProgAgent yields higher pass rates than the evaluated baselines of comparable size across four benchmarks. On LOCA-bench, ProgAgent-14B achieves a 30.7% pass rate at a 64k-token environment state, where these baselines score zero. Cross-harness experiments show that PTC-trained models outperform DTC-trained models under both evaluation interfaces, suggesting that the benefits of PTC training extend beyond the interface.

Linear attention promises the efficiency needed for long-context language modelling, yet its fixed-size recurrent state must simultaneously support ongoing computation and preserve information from an ever-growing history. This coupling has reinforced a prevailing view that the state's memory capacity defines the model's long-context capability. We challenge this presumed ceiling: recurrent computation need not bear the full burden of long-term storage. We introduce NoteBank, a sparse episodic memory that archives selected text before future queries are known, uses the backbone's native features for writing and addressing, and reads retrieved evidence through its native forward computation via replay. This training-free, plug-and-play interface requires no external embedding model, separately trained retriever, or separate neural reader. At one million tokens, it raises Gated DeltaNet's mean accuracy across six RULER retrieval tasks from 1.7% to 70.3%, with no observed significant retrieval degradation from 4K to 1M tokens. This recovery adds leqslant0.025% to analytical FLOPs, with persistent context storage approximately 1/2,600 that of a Transformer's full MHA KV cache at matched width, depth, and precision. Experiments extend from three 1.3B recurrent backbones to Mamba-2-8B and post-trained Qwen3.5-9B, demonstrating gains on LongBench v1/v2 and MemoryAgentBench. Together, these results show that sparse evidence storage and native replay can substantially expand the long-context capability of pretrained linear attention models while retaining the efficiency of recurrent computation.

Generative discovery in science requires coupling high-level scientific reasoning with the ability to generate samples respecting the structure of a design space. Yet these capabilities remain largely separated: foundation LLMs encode broad chemical and biological knowledge, while domain-specific flow- and diffusion-based generators capture the priors needed to produce valid sequences and geometries. We study how to align such scientific generators to high-level natural-language discovery goals without requiring an explicit proxy reward, which is often unavailable or poorly captured by learned predictors in experimental sciences. We introduce LLM-to-Flow Distillation (LFD), where a frozen LLM acts as an adaptive teacher for a domain-specific generator. Given a goal such as 'generate peptides with promising anticancer activity', the teacher constructs a curriculum of intermediate sub-goals aware of the flow's current capabilities, and progressively directs them toward the original discovery goal. At each round, the LLM judges sampled candidates under the current objective. These judgments induce preferences that are iteratively distilled into the generator through DPO. This procedure provides dense learning signals even when the initial model places negligible probability on designs satisfying the final goal, enabling adaptation toward poorly covered, out-of-distribution regions. Experiments on therapeutic peptides and small-molecule structure design demonstrate that LFD enables a new form of generative control: aligning arbitrary domain-specific scientific generators to scientific discovery tasks expressed via high-level natural-language goals.

Sparse attention reduces the KV-cache traffic that dominates long-context LLM decoding, but practical methods largely operate at coarse granularity. Token-level selection offers finer control, yet presents two challenges: low-dimensional scoring may lose query--key structure needed for accurate ranking, while gathering scattered selected tokens can offset the resulting latency benefit. We introduce Joint Subspace Sparse Attention (JSSA) to address both challenges. For selection quality, JSSA learns a shared low-dimensional query--key subspace by minimizing token-score preservation error. We upper-bound the resulting non-convex objective with a tractable marginal objective that admits a closed-form spectral solution; direct refinement of the exact objective provides no measurable improvement in the tested settings. For execution efficiency, we develop a gather-free token-level indexed-attention kernel that supports distinct per-KV-head selections under the standard grouped-query attention (GQA) KV-cache layout. Across standard GQA, hybrid sliding-window/global attention, and Multi-head Latent Attention (MLA) models, JSSA consistently remains close to full-KV accuracy while avoiding the substantial generation-length inflation and architecture-dependent failures observed with competing methods. On an NVIDIA B200 GPU, JSSA achieves up to 6.43times the decoding throughput of full-KV attention at 128K context. JSSA achieves reasoning accuracy comparable to model-native DSA without model training and reaches 96--99\% of its decoding throughput even at 700B scale large models.

Rebecca Rubin / Variety : Netflix Chief Content Officer Bela Bajaria is leaving the company and will be replaced by former Disney and Fox executive Peter Rice, who starts October 26 —  In a major executive shakeup, Bela Bajaria is out at Netflix after 10 years at the streamer, most recently serving as its chief content officer.

Sarah Perez / TechCrunch : Bluesky expands Attie, its AI tool for building custom feeds, with a “social vibe coding” feature to create sites, games, quizzes, and more on the AT Protocol —  Open social network Bluesky is once again expanding beyond its flagship app.  At TechCrunch Disrupt 2026 on Thursday …

Reuters : Sources: some Anthropic investors discussed a ~$1.5T valuation for its IPO, down from ~$2T they had expected, after disclosures showed slowing Q3 revenue growth —  Some current Anthropic investors and prospective investors in its upcoming initial public offering are weighing a lower valuation …

Nieman Lab : Experts say Meta's new terms of use will allow scraping of websites, including news sites, “associated with” business accounts on its platforms, with no opt out —  Starting October 30, business accounts will consent to Meta scraping their websites if they continue using Facebook, Instagram, or WhatsApp.

Jaspreet Singh / Reuters : Thoma Bravo-owned cybersecurity company Proofpoint plans to buy data security firm Varonis Systems in an all-cash deal valued at about $6.3B —  Thoma Bravo-owned cybersecurity company Proofpoint will buy data security firm Varonis Systems (VRNS.O) in all-cash deal valued at about $6.3 billion, the companies said on Thursday.

Kate Sullivan / Bloomberg : A US official says Trump didn't order any limitations on Ukraine's access to SpaceX's Starshield; Elon Musk said any limits are ultimately decided by POTUS —  President Donald Trump didn't order any limitations on Ukraine's access to SpaceX's Starshield satellite technology, according to a senior US official.

Nick Cammarata / Goodfire : A researcher says modern neural networks are far more interpretable than believed, with most neurons understandable in models up to trillions of parameters —  The cause of polysemanticity in early model organisms wasn't superposition, but several ordinary, mostly fixable issues.

Reed Albergotti / Semafor : Perplexity open sources SPACE, its AI agent sandbox, and says partners including CoreWeave, Akamai, and Crusoe agreed to offer SPACE to customers —  THE SCOOP  —  Perplexity is giving away the software it has been using to combat rogue AI agents, the company told Semafor Thursday.

Ben Penn / Bloomberg Law : Sources: top DOJ officials halt career prosecutors' Biden-era investigation of Meta over facilitating illegal drug sales and are weighing more lenient options —  Top Justice Department officials have sidelined career prosecutors' long-running grand jury investigation of Meta Platforms …

Andrew Deck / Nieman Lab : Google says it has 400+ publications in its News AI pilot, which tests features like overviews; participants include Financial Times, CNN, and The Texas Tribune —  In December, Google launched its News AI pilot program, which licenses content from select publishers and uses it to test new AI features in Google News.

CTech : Two workers have been killed and another seriously injured at Intel's Kiryat Gat fab in Israel in an explosion resulting from a gas cylinder malfunctioning —  A 49-year-old man died Thursday from injuries sustained in Wednesday's blast, which also killed a gas technician and left a third worker in serious condition.

Andrew J. Hawkins / The Verge : Volvo says it will use Nvidia's Hyperion platform to launch a Level 2++ hands-free, point-to-point automated driving system by 2030 —  The system will be similar to Tesla's Full Self-Driving in that it will enable hands-free, point-to-point partially automated driving on local roads.

Joseph Cox / 404 Media : Memo: Accenture has told many employees, some of whom had been spending as much as $1,000 monthly, that it will now only spend $29 a month per person on Claude —  Some employees who spend as much as $1,000 on Claude usage each month are scrambling to find ways to maintain access to the tool.

Anthropic : Anthropic says it is giving employee-like access to METR as part of its pledge to partner with evaluators for assessment of its models and safety practices —  Last month, we committed to stronger independent, third-party assessment of our models and safety practices.

Ryan Lawler / Axios : Sunbound, which makes software for handling billing, insurance claims, and payments for senior living operators, raised a $24M Series A led by TTV Capital —  Sunbound, which handles billing, insurance claims and payments for senior living operators, has raised $24 million in Series A funding …

美国银行（Bank of America）CEO Brian Moynihan 透露该公司雇佣了千名合同工帮助修复 Anthropic Mythos 等 AI 工具发现的软件漏洞。他表示：“我们在第二季度聘请了 1000 名合同工协助进行修复工作。除了原本负责该领域的 3000 名员外，这些合同工至今仍留在公司内，继续提供协助。”他向分析师透露，美国银行第三季度在软件修复以及与反洗钱合规相关的监管问题上投入了数亿美元以上的资金。这是美国大型企业首次披露其为保护软件系统免受 Mythos 影响所投入的人力和资金规模。

加拿大多伦多大学公民实验室发表研究报告，以色列公司 Discover-Tech 通过数字广告在半岛电视台的内部设备上安装了间谍软件。这是首次利用数字广告实施的零点击间谍软件攻击。Discover-Tech 运营着名为 Controlium 的平台，它表面上是一家合法的广告公司，但实际上是以色列间谍软件供应商的合作伙伴。Controlium 能在广告加载的瞬间，对设备进行去匿名化处理、提取设备指纹，自动植入恶意软件。对 Discover-Tech 四十亿条竞价记录的分析显示，在四个月时间里，该公司的广告出现在数十个国家、近 20 万个不同的网站及 Android/iOS 应用中。公民实验室称，名为 DARK CARTOGRAPHER 的组织正利用 Discover-Tech 的基础设施，通过挖掘多家移动运营商系统中的漏洞，将设备的广告标识符与电话号码进行关联，对互联网用户实施大规模的去匿名化。Google 等已采取行动，切断了 Discover-Tech 接入数字广告生态系统的渠道。

Canonical 释出了代号为 Stonking Stingray 的 Ubuntu 26.10，这是一个短期支持版本，仅支持九个月。Ubuntu 26.10 的主要变化包括：用 Rust 语言开发的版本替代了 C 语言开发的系统组件，后量子密码学软件包签名，扩展对基于 TPM 的全盘加密支持，企业级身份验证与 Microsoft Entra ID 的更深度集成，Linux kernel 7.3，新 RISC-V 桌面镜像，GNOME 51 桌面环境，可选启用的本地处理语音转文字方案 Myna，等等。

代表美国科技巨头亚马逊、Google 和 Meta 的行业组织 CCIA Europe 委托进行的一项研究声称，欧洲如果推行数据主权法案 Cloud and AI Development Act（CADA） 会在十年内带来 1150 亿欧元的成本，是欧盟委员会预估的最高值的六倍多。其中 800 亿欧元的成本是假设数字主权法扩大至私营部门关键服务四分之一的云工作负载。报告特别针对了 CADA 中的第 31 条，称其可能允许欧盟委员会将数字主权要求扩大到私营部门的关键服务领域。CCIA 认为强迫企业放弃现有的云基础设施，可能会引发昂贵的软件重开发，其中代码重写和架构重构占预计成本的 69%。

Google 现在会自动验证用户的手机号码，将其关联至用户 Android 设备上登录的所有 Google 账户，全程不会弹出明显确认提示。该功能在 Google 设置中被称为电话号码验证(Phone number verification)，会将用户当前使用的手机号码关联至同一台设备上登录的所有 Google 账户——无论是辅助账户、工作账户，还是其它临时账户。用户通过社交媒体报告收到了 [email protected] 的确认邮件，通知其手机号码已通过验证。调查显示，Google 支持页面于 2025 年 8 月至 10 月间进行了更新，默认启用电话号码的验证功能。用户可通过：设置——Google 或 Google 服务与偏好设置——所有服务——隐私与安全——电话号码验证，关闭该自动电话验证功能。

多项调查数据显示，美国人的社交活动过去三十年大幅减少。American Time Use Survey 的调查数据显示，相比二十年前，美国人独处的时间更多，与朋友相处的时间更少。American Perspectives Survey 的调查发现，有五个或更多密友的美国人比例下降了 25 个百分点，从 63% 降至 38%。Data for Progress 的调查发现，从 1995 年到 2026 年，美国人参与各项关键社交活动的比例出现了显著下降，降幅均达到两位数：去上课或参加研讨会（-22）、在家招待客人（-21）、去教堂或其它宗教场所（-20）、外出野餐（-19）、参加社团聚会（-19）、打牌（-19）、去餐馆吃早餐（-13）以及从事志愿服务（-12）。与此同时使用互联网（+90）、在家玩游戏（+22）、参加健身课程（+14）、音乐会（+13），因个人原因乘飞机出行（+5）。

科学家利用南非 MeerKAT 射电望远镜和韦伯望远镜，捕获并定位了迄今最远的快速射电暴 FRB 20240304B，其源头位于大爆炸后约 30 亿年的星系。该信号在宇宙中走了 100 多亿年，距离是此前 FRB 纪录的两倍多。快速射电暴（FRB）从第一次被发现以来，一直令天文学家困惑不解。该现象是持续时间数毫秒、来自银河系外的射电脉冲，其物理起源神秘，却能量极大。尽管付出了很多努力，科学家也一直在发展相关理论，但迄今仍未完全厘清 FRB 的来源。目前磁星是部分来源的候选者。此次快速射电暴只持续几毫秒，同样释放巨大能量。2024 年 3 月 4 日，MeerTRAP 项目用 MeerKAT 先抓到爆发，天文学家再用韦伯近红外相机和光谱仪找它“老家”。光谱给出红移约 2.148，这对应宇宙还很年轻的时候，以往地面大望远镜看不到这个宿主，因为它又小又暗，但韦伯的红外能力可以把它锁定。确定位置后，天文学家发现宿主不像多数 FRB 所在的成熟大星系，而是一个低质量、贫金属、正在猛造恒星的矮星系。这类环境年轻、造星活跃，更支持“磁星起源”：大质量恒星变成超新星后留下强磁场中子星，短时间内就能产生爆发。

珊瑚白化是指温暖海水带来的热应激导致珊瑚排出与其共生的甲藻；这是全球珊瑚礁衰退的一个主要原因。研究人员如今证明，健康珊瑚与白化珊瑚之间的短暂物理接触可将藻类重新引入白化珊瑚，使其完全恢复。仅需一周——有时甚至短至三天——就足以触发这一转变。藻类的转移是通过珊瑚之间的组织融合实现的；因此该方法仅对来自相同基因背景（或基因型）的珊瑚有效。研究人员发现，在相同基因型和混合基因型的配对中，共生藻都会在接触时通过珊瑚虫的摄取发生转移；他们指出，下一步的工作是观察这些基于接触的恢复策略能否在野外应用于自然珊瑚礁这样的规模。

微软的企业安全软件 Microsoft Defender 正式支持包括 Ubuntu、Debian、Fedora 和 RHEL 在内的 Linux 发行版，其功能包括防病毒保护、端点检测与响应、漏洞管理、威胁搜寻以及设备隔离。Microsoft Defender 是企业安全套件，集成在 Windows 11 中的免费安全软件叫 Windows Defender，Windows Defender 是 Microsoft Defender 中一个功能有限的防病毒组件。使用 Microsoft Defender 需要用户订阅微软的服务，订阅 Microsoft Defender for Endpoint Plan 2 或者 Microsoft 365 E5 或 E7。

堪萨斯州、亚利桑那州和蒙大拿州等美国八个州对 OpenAI 提起诉讼，指控 ChatGPT 诱导儿童自我伤害。八个州要求 OpenAI 改变其产品与年轻用户互动的方式，没收由此产生的利润。诉讼指出 ChatGPT 的设计初衷是通过表现得随和、善于互动去赢得用户信任。但诉讼称聊天机器人会强化妄想、鼓励危险行为，协助用户策划诸如大规模枪击之类的暴力行为。诉讼指控 OpenAI 在 ChatGPT 的安全性上误导用户，违反了各州的消费者保护法。它们还指控 OpenAI 因危及当地居民的健康而构成了公害。诉讼要求法院下令强制 OpenAI 对其产品 ChatGPT 的工作方式进行整改。其中一项要求是除非 OpenAI 能确保其 AI 模型符合人类价值观，否则应禁止其开发具备自我完善能力的 AI 模型。这些州还寻求对 OpenAI 违反州法的行为处以民事罚款，要求对损害赔偿。

一项新研究估计，2023 年有 1140 万人患上痴呆症，到 2050 年预计这一数字将增加一倍以上，新增痴呆症人数将达到 2470 万。研究发现女性的痴呆症发病率高于男性，且低中收入国家的发病率远高于高收入国家。在校正年龄因素后，女性发病为每千人年 10.6 例，男性为 8.6 例。低中收入国家的发病率为每千人年 13.6 例，而高收入国家为 8.3 例。目前全世界有逾 5700 万人受痴呆症困扰，到 2050 年这一数字将达到 1.5 亿。每年新增 2400 万痴呆症病例，将给个人、家庭、卫生系统及经济带来巨大压力。

Rockstar 尚未正式宣布，但有报道称它计划于 2027 年第一季度发布 GTA 6 的 PC 版本，与主机版本仅仅相隔几个月。GTA 6 将于 2026 年 11 月发布主机版本。据彭博报道，在 GTA 6 开发过程中 Rockstar 员工总数从开发 Red Dead Redemption II 的 2,441 人增长到 4,711 人。GTA 6 制作预算据称超过 20 亿美元，提前释出 PC 版本旨在加快回收制作费用。上一代 GTA 5 的主机版本和 PC 版本相隔 1 年 7 个月，而 GTA 5 的主要受众是主机玩家，PC 玩家占少数。如今 PC 玩家的规模已经显著扩大。加快发布 PC 版本也有助于扩大 GTA 6 的受众规模。

如何吸引 Chrome 的用户改用其它浏览器？Chrome 是目前市场份额最高的浏览器。苹果推出了一个广告页（无中文）：Switch to Safari。苹果列举了一系列 Safari 的优点，称在 Mac 设备上，相比 Chrome，使用 Safari 播放流视频，续航时间最长可多出 5 个小时；常用网站的页面加载速度更快；单次充电即可全天播放流视频；动画内容渲染性能更出色；复杂 Web 应用 JavaScript 运行速度更快；默认阻止第三方 Cookie 追踪；屏蔽恼人的干扰内容，等等。苹果称，Safari 在速度、电池续航以及隐私上都超过 Chrome。

Linux 作者 Linus Torvalds 在维护者峰会上透露他在大模型帮助下开发了一系列补丁解决一个长期存在的问题。现在这组 10 个补丁已递交接受审查。这些补丁主要增强 folio 的查找机制：通过识别处于热数据状态的 folio，设置一个“延迟释放该 folio 直至 RCU 周期结束”的标志。随后利用 RCU 延迟在 RCU 保护下进行常规的 folio 查找，同时在 RCU 保护下直接进行实际的 folio 数据访问...补丁在测试中性能实现了数量级的提升，但已发现存在竞态条件，因此需要进一步完善。

荷兰税务与海关总署放弃了部署微软 Microsoft 365 公有云的计划，改为采用本地部署/开源替代方案。根据 Linux 基金会在欧洲开源峰会上发布的报告，94% 的欧盟组织认为数字主权重要，其中 61% 认为非常重要。除了税务机关，荷兰还在探索全面从 Windows 和私有软件转向 Linux 和开源软件。荷兰原计划于 2025 年转向 Microsoft 365，计划在 2026 年底前完成全面部署。但 Dutch Advisory Council on ICT Assessment 在今年 6 月建议叫停部署。政府于 7 月暂停了该计划，现在决定转向开源方案。税务与海关总署目前还不会全面转向开源替代，而是暂时保留现有的微软软件。