OrangeBot.AI Digest — 2026-10-31
89 headlines across 8 sources, aggregated for this day.
Hacker News(15)
- Anthropic blocks users for anything touching training larger models (twitter.com)
- HackerTube: YouTube built from videos recommended by HN (karpathy.bearblog.dev)
- TCP Use-After-Free: CVE-2026-98432 / CVSS 9.8 (atlasum.net)
- GPT-5.6 Sol can beat Factorio on peaceful mode (ryanmadden.net)
- UK Green party drops longstanding opposition to nuclear power (www.theguardian.com)
- Jepsen: ParticleDB 1.1.4 (jepsen.io)
- Casio Watches of the 1980s (www.casio.com)
- Malware uses LLMs to rank victims by how much they're worth (blog.talosintelligence.com)
- Simple circuits can create complex behaviors (aatishb.com)
- Prospero: A new coding harness for a more civilized age (nullprogram.com)
- Do the math: How we took over ChatGPT accounts (www.parameter.ai)
- Molly (Signal fork) has not been updated since 3 months and is vulnerable (piefed.world)
- Serbia just elected the world's brainiest parliament (www.science.org)
- AI, Lean, and the Future of Formal Verification, with Lean Creator Leo de Moura (thedataexchange.media)
- Digital minimalism (vitalik.eth.limo)
GitHub Trending(14)
- thedotmack / claude-mem
- mattpocock / skills
- Donchitos / Claude-Code-Game-Studios
- pingdotgg / t3code
- Renhuai123 / ziwei-doushu
- milind-soni / OpenMausBot
- pbakaus / impeccable
- obra / superpowers
- lightpanda-io / browser
- HKUDS / DeepTutor
- Orkas-AI / Orkas
- THU-MAIC / OpenMAIC
- bregman-arie / devops-exercises
- vorssaint / vorssaint-utils
Product Hunt(15)
- LumaFold: Screen Light
Makeup Mirror & Color Effects for iPhone Duo
- WinnerAdSpy
Turn competitor ad insights into winning app campaigns
- SpaceO
Give AI agents their own screen on your Mac
- GitHub Copilot Fusion
GitHub's AI coding agent that picks and combines models
- BaseTab AI
Turn your new tab into an AI-powered workspace
- Backenly
The autonomous backend for AI coding agents
- AC Bridge
Your smart lights show when Claude Code or Codex needs you.
- Pluto Door
SSH client with an AI agent and a desktop for every server
- Agnos
Clean documents for AI, and see what was removed
- SKI
Voice-first coding harness
- AwaitIO
Write a TypeScript function, get an admin screen
- Mo
A little character on your Mac who does 232 tasks for you
- Specsight
Product observability for teams shipping fast with AI
- telemetry.dev
LLM observability for AI agents
- Seqora
Design your schema on a canvas. Seqora writes the SQL.
Hugging Face(15)
- Look Only When Needed: Certificate-Guided Visual Routing for Efficient Diffusion VLMs
Diffusion vision-language models (DVLMs) process their entire input at every denoising step. Since visual tokens make up 89-97% of this input on the benchmarks we study, most of the decoding computation is spent repeatedly processing the same image. Visual token pruning methods partially relieve this burden by removing part of the visual tokens, but they lose the removed information and still process the remaining image at every step. To measure how often the image is actually needed, we compare LLaDA-V's predictions with and without the image at every decoding step and find that removing the image changes a token being unmasked at only 22.8% of steps on average and at as few as 7.9%. We therefore propose LOWN, a decoding method that decides at every step whether to process the image and how many tokens to unmask. LOWN predicts the masked tokens without the image and processes the image only when this prediction disagrees with the last image-conditioned prediction. Agreement certifies the earlier prediction for reuse, and the number of certified tokens sets how many are unmasked in that step. We further observe that the need for the image differs across benchmarks, so LOWN's adaptive decision is needed rather than a fixed schedule. Without training or a tuned threshold, LOWN retains 99.6% of LLaDA-V's score on ten benchmarks at 35.3% of the decoding FLOPs on average. LOWN also generalizes across LLaDA-based DVLMs, composes orthogonally with visual token pruning, and runs under key-value caching for further efficiency gains. Code and project page are available at https://hyeon9698.github.io/LOWN/
- Youtu-Parsing-Omni: One Encoder, One Schema, All Modalities
We present Youtu-Parsing-Omni, an omni-modal parser that converts documents, images, audio, and video into structured records grounded in space and time, providing a perception layer for agents that read, locate, and cite evidence across modalities. Its design rests on one encoder for every modality and one schema for every task. Youtu-Omni-Encoder, initialized from a pre-trained language model to strengthen text reading, encodes all modalities with shared weights in a unified space-time coordinate system, and a few fusion layers let video frames attend to concurrent audio, bringing cross-modal fusion and temporal alignment into perception itself. OmniSchema unifies the outputs of all ten parsing tasks, allowing a single self-evolving post-training procedure, OmniSchema-Aware On-Policy Distillation (OSAD), to provide a shared token-role-based objective across tasks. In each round, the model is trained on its own rollouts against token-level targets from a reference-conditioned copy of itself, which is then refreshed with the trained weights, so that the parser and its supervision co-evolve. With a single 5 B-parameter checkpoint, Youtu-Parsing-Omni achieves state-of-the-art performance on OmniDocBench v1.6 (96.96 Overall) and leads all open-weight models on OmniParsingBench (75.08 Avg.), with its advantage concentrated on audio and video. It also achieves the best overall chart-parsing score (95.02) and the highest exact chemical-fingerprint agreement among compared systems, and matches a dedicated optical-music model in character-level transcription of typeset sheet music while staying far ahead of general-purpose systems. Code and model weights are publicly available.
- A-RSI: Adaptive Task Refresh for Zero-Data Multimodal Self-Improvement
Recursive self-improvement requires models not only to acquire useful training experience, but also to decide when their current experience has ceased to be useful. Existing methods often focus on what experience to generate next, while paying less attention to when fresh experience should be introduced. Many also construct training targets through self-consistency or majority voting, making supervision depend on the model's own uncertain predictions. We address these two questions separately: training dynamics determine when to acquire new experience, while the answers to new tasks remain independently checkable. A-RSI monitors reward variation and accuracy in training logs to refresh a task pool when its learning signal declines, without consulting validation or test scores. A static library of model-authored programs produces multimodal questions and derives their reference answers; we specify the task-generation contract, automatically verify chosen properties of the reward and refresh rules, and separately audit the generated answers. The training loop uses no external training examples or human-written answer labels. On the primary eleven-benchmark suite, A-RSI raises the performance average from 57.77 to 61.05 with 800 fresh tasks.
- Proving and Improving the Platonic Representation Hypothesis with JEPAs
Whether models trained on different modalities converge to the same representation of the world is central to AI. The Platonic Representation Hypothesis (PRH) conjectures that they do as they scale, but without a proof it offers neither a guarantee nor a recipe. We turn the hypothesis into a theorem. First, we prove a Platonic Representation Theorem: Joint-Embedding Predictive Architectures (JEPAs) trained separately on any number of modalities learn the same representation up to one rotation whenever their positive pairs are two observations of one scene--good views--and their misalignment is bounded otherwise. Second, we show that next-token prediction is a JEPA with bad views, parts of a single document, which explains why language models align with vision only through scale. Third, we validate the theory from 145k captions to the texts of 940M web images, and uncover a scaling law of alignment that predicts unseen runs within seed noise. Good views need 42 times fewer training samples than next-token prediction on the same corpus to reach the same alignment, and a 109M-parameter text JEPA aligns with DINOv2 and CLIP's image encoders better than every language model we evaluated, up to 72B parameters, and than CLIP's text encoder: next-token prediction is a remarkably inefficient route to platonic representations. As a showcase, one rotation turns separately trained encoders into an image captioner (CIDEr .838 on COCO). Even that rotation needs no pair: on the latest pair-free benchmark, our text JEPA beats every zero-shot accuracy reported with language models up to 8B parameters, e.g., 64\% against at most 56\% top-5 on ImageNet-100. Positive views, and not scale alone, are thus the lever to build platonic representations.
- L3G-World: Real-Time Infinite World Generation with Latent 3D Gaussian Memory
Interactive world models generate visual content conditioned on user actions, yet maintaining spatial consistency over long-horizon exploration in real time remains challenging. Existing approaches rely on either temporal context or explicit 3D memory, both of which scale poorly because memory grows with the length of the generated stream. Moreover, explicit 3D memories condition latent video generators by re-encoding RGB renderings, which precludes real-time generation. In this paper, we introduce a Latent 3D Gaussian Memory (L3G) that combines the efficiency of feed-forward reconstruction with a compact, rapidly renderable scene representation. Building on this memory, we propose L3G-World, in which a memory updater and a video diffusion transformer (DiT) alternate autoregressively to generate spatially consistent worlds. The memory updater spawns Gaussians in newly observed regions and refines existing ones, so that the memory grows with scene extent rather than stream length. Each Gaussian further stores a VAE latent feature, allowing the memory to be rendered directly into the DiT's latent space without VAE re-encoding. Experiments show that L3G-World achieves state-of-the-art generation quality and long-horizon spatial consistency, with up to 38.6times fewer 3D primitives, 3.2times faster memory updates, and 359times faster memory rendering relative to the most efficient prior 3D-memory baseline for each metric, enabling real-time world generation at 9 FPS.
- Read, Allocate, Route: Role-Coupled Tokenization for Pixel-Space Generative Transformers
Pixel-space generative Transformers typically use one token per image patch, so capturing finer detail requires more tokens and makes global attention increasingly expensive. We observe that finer detail does not necessarily require more global communication: low-frequency structure benefits from global exchange, while high-frequency detail can be modeled locally. Based on this insight, we introduce SpecSlot, a pixel-space rectified-flow Transformer that decouples output detail from the length of the global attention sequence. SpecSlot expands each spatial anchor into multiple slots, each assigned a disjoint band of DCT coefficients; this separates anchor density from spectral subdivision. Only the lowest-frequency slot from each anchor participates in global attention, while all slots interact locally. This allows the model to increase spectral detail without enlarging the global sequence. A spatial head further adds a local correction to the spectral prediction, while the generative state, target, and loss remain in pixel space. On ImageNet-256, SpecSlot achieves a comparable final FID to a FLOPs-matched JiT control while converging faster. On ImageNet-512, it improves FID by 2.84 over the control with 3.9% fewer FLOPs, and comes within only 0.69 FID of JiT-B/16 at 38.2% lower cost. On text-to-image generation, SpecSlot surpasses JiT-B/8 on DPG-Bench by 1.06 at 40.6% lower forward cost.
- WorkFactory: Turning Professional Intents into Executable Work for Scalable Agent Training
Scaling supervised training for working agents requires diverse professional tasks that yield reliable expert demonstrations. However, missing or inconsistent task information can obstruct teacher execution and weaken the resulting supervision. We propose WorkFactory, a scalable framework for constructing coherent and executable professional tasks. Starting from professional intents, information-level planning builds an information support graph to guide task instantiation, explicitly representing the dependencies and locations of information required for the intended outputs. Probe executions then identify support failures and guide repairs while preserving the intended work. We construct 112,686 tasks spanning 612 occupations and use the collected teacher trajectories to fine-tune Qwen models at 4B, 27B, and 35B-A3B scales, yielding the WorkPilot family. All three models improve over their respective base models across four professional-work benchmarks, with WorkPilot-35B-A3B improving from 879 to 1565 Elo on GDPval-AA v2.1. Scaling analyses reveal model-dependent diminishing returns, with execution length first increasing and then decreasing. Controlled component analyses show that planning and repair reduce probe-diagnosed task environment defects and improve downstream performance. The data and models are available at https://github.com/IQuestLab/WorkFactory.
- A Broader Look into MLLM's Retrieval Heads
Locating relevant evidence in long inputs is a core capability of Multimodal Large Language Models. Prior studies attribute this capability to retrieval heads, a sparse set of attention heads that transfer information from the context to the output. However, these analyses are confined to a limited set of modalities and the corresponding models. Moreover, how retrieval heads can be leveraged to enhance the model calls for further exploration. In this paper, we ask whether this mechanism expands together with the input modalities of the model. Taking audio-visual input as a newly added modality, we construct paired inputs in which the same answer is either spoken or shown on screen. With the evidence content fixed, retrieval heads remain sparse in the new modality and are shared to a large extent across modalities. Across 18 language, vision-language, and omni-modal checkpoints, omni-modal models largely preserve the retrieval heads of their backbones. Ablating these heads eliminates nearly all audio-visual retrieval. The heads shared across modalities make up fewer than 2\% of all heads, yet ablating them also degrades document understanding, spatial perception, and visual mathematics. Building on these findings, we turn retrieval heads into a target of optimization by training their attention with a ranking objective. This raises the page Recall@1 of Qwen3-VL-8B on MMDocIR by 6.03 points and improves document retrieval across three model families. Functional attention heads can thus serve both as a lens on multimodal models and as a target for improving them.
- VibeGame: Prompt-to-Game Development with AI-Native Engine and Self-Evolving Adversarial Agent Team
Large language model agents have enabled software development through natural-language interaction, but extending this paradigm to end-to-end game development remains challenging. Building a playable game requires coordinating code, scenes, assets, animations, physics, and configurations, while allowing developers to validate their integrated behavior through interactive testing. We introduce VibeGame, a framework for creating complete 2D game projects from natural-language prompts and optional reference images. Its AI-native game engine represents project state as schema-validated structured text and provides frame-synchronous runtime control, enabling agents to inspect game states, execute actions, and verify behavior at their own pace. Built upon this interface, we design an Adversarial Agent Team of eight specialized agents to separate implementation from review, so that problems are found and fixed through multi-round loops. For each conducted and accepted project, we further introduce a training-free self-evolution mechanism to distill it into reusable skeletons, modules, and collaboration contracts for future development. We evaluate prompt-assets-to-game (PA2G), where agents are given an asset library, and prompt-to-game (P2G), where agents generate their own art. On a GameCraft-Bench subset, VibeGame in P2G consistently outperforms a range of agent harnesses in either P2G or PA2G with the same underlying model, including GPT-6.1-Sol.
- Your Benchmark Is an Empirical Measure Over Difficulty
A benchmark is not merely a collection of test questions; it naturally induces an empirical measure over difficulty. In this work, we develop this measure-based view of benchmark evaluation. This perspective allows us to identify and formalize two central failure modes: measure mismatch, where the benchmark difficulty distribution differs from the intended difficulty distribution, and coverage gaps, where uncovered regions along the difficulty axis prevent benchmarks from detecting meaningful localized training progress. We propose two diagnostic metrics that capture these two failure modes and provide rigorous theoretical guarantees. Empirically, we analyze 25 popular benchmarks spanning 7 domains using carefully estimated item difficulties. Our results show that many benchmarks implicitly overweight certain difficulty regions, so their leaderboard conclusions should be interpreted with caution; others leave substantial portions of the difficulty axis uncovered, limiting their ability to monitor the model improvement process. Finally, we turn diagnosis into action by introducing BenchFill, a simple difficulty-targeted item rewriting algorithm that effectively reduces local coverage gaps.
- HairDeck: Animation-Ready Stylized Hair Cards from a Single Image
The hair of stylized characters in games and avatars is built from hair cards: artist-placed ribbons whose geometry is the visible hair, animated by spring bones. Existing image-based methods produce realistic hair or a single fused mesh, while CHARM, the only generator of stylized hair cards, decodes a hairstyle autoregressively as a long sequence of control points. Observing that a card hairstyle is an unordered set of artist-defined primitives, we propose HairDeck, which generates it as one: each card is a single token, and a set diffusion transformer denoises all cards together. From a single image, HairDeck generates the cards and the spring-bone chains that drive them, yielding an animatable asset that exports to ready-to-use downstream formats. On CHARM's own test set, it lowers Chamfer distance by 44% and card-orientation error by more than half while running 33 times faster, and from single images it outperforms CHARM, image-to-3D pipelines and strand-based methods. Project page: https://hyzcluster.github.io/HairDeck
- Co-Design Bench: Multimodal Collaborative Coding Evaluation for Agentic Models
Evaluating human-AI co-design requires testing whether coding agents can translate real users' multimodal requests into working application revisions. We introduce Co-Design Bench, a reference-free benchmark of 273 tasks curated from 37,463 deduplicated queries across 26,314 real user coding sessions. Its two tracks, Multimodal Editing (ME) and Repository Editing (RE), cover localized edits with optional image, audio, video, or document inputs and coordinated repository revisions. Tasks span three capability dimensions: Grounded Understanding, UI Transformation, and Functional Implementation. The tasks pair normalized requests and runnable pre-edit applications with 2,331 hidden expert-authored acceptance criteria. Our judge-then-refute protocol grades criteria using source and browser evidence, challenges positive verdicts with a separate LLM verifier, and audits visual and interaction claims without gold code or target screenshots. Under a shared agentic harness, the strongest of ten model endpoints passes an average of 69.9% of criteria per task but fully satisfies only 26.4% of tasks. All ten endpoints achieve lower criterion completion on repository edits spanning three or four change types than on those spanning one or two. Co-Design Bench provides a testbed for studying the gap between partial requirement satisfaction and complete task success under agentic execution for real-user application revisions.
- Intrinsic Distribution Matching
One-step and few-step text-to-image generation must accurately render the details specified in the prompt despite limited opportunities for refinement. Distribution matching distillation (DMD) uses score differences to bring the generated distribution closer to a reference distribution. Following prior work, we estimate this difference with internal velocity branches, avoiding a separate fake-score network. We ask whether an explicit moment constraint improves this learned score supervision. A decomposition of the reverse-KL field into a Gaussian moment component and a non-Gaussian residual identifies statistics that can be constrained explicitly. We introduce Intrinsic Distribution Matching (IDM), which retains the full DMD correction and matches feature means and per-coordinate standard deviations using the generator's exponential-moving-average (EMA) copy, requiring no external feature encoder. Experiments use full-parameter and low-rank adaptation (LoRA) tuning on Z-Image-Turbo 6B and Qwen-Image 20B. At two network evaluations, IDM reaches 94.25% and 90.20% on Chinese and English LongText-Bench, the highest scores among the compared methods at this budget. Controlled ablations show gains from the added moment loss, while generator EMA features perform competitively with an external DINOv2 encoder across the five reported metrics. Code is available at https://github.com/inclusionAI/IDM.
- WeVisCut: The Whole Editing Room in One Agent From Narration to Long-Form Video
Producing long-form videos from narration requires interdependent decisions that remain coherent across shots. Existing agentic editors often use multi-stage workflows driven by general models, leaving the quality-cost trade-off of compact, task-adapted policies unclear. We address this gap with WeVisCut, a long-form script-to-video system that connects tool execution with policy learning. Its agent harness maintains shared production state so that tool feedback informs subsequent decisions and repairs. Because finished videos do not expose these decisions, our four-stage data pipeline combines editing inversion with verified replay to turn recovered editing intent into executable supervision. Applied to 3,750 finished videos, this pipeline yields over 130K training examples. We use this supervision to train a single 9B policy that coordinates production and recovers from failures within the same harness. Moreover, we introduce a 100-task benchmark and five metrics assessing editing quality and delivery reliability. Under a shared tool environment, WeVisCut substantially outperforms its base model and achieves comparable performance to Claude Opus 5 while using nearly 33x fewer generated tokens per successful output minute. It also outperforms task-matched native editors on our proposed metrics.
- IntReal: A Real-World Dataset and Benchmark for Multimodal Interleaved Generation
Humans routinely consume multimodal content, where textual and visual information is naturally interleaved. Although Unified Multimodal Models (UMMs) can jointly generate text and images, multimodal interleaved generation remains largely underexplored. Progress in this area is hindered by two key limitations: (1) the lack of well-defined interleaved generation tasks and corresponding training data, and (2) the absence of fine-grained evaluation protocols that can reliably diagnose model performance across different capability dimensions. To this end, we present IntReal, comprising two complementary components: IntReal-100K, a real-world dataset of multimodal interleaved samples spanning four major categories, and the IntReal Evaluation Suite, which evaluates model performance across seven complementary dimensions using specialized expert models, enabling coarse-to-fine diagnosis. Extensive experiments validate our evaluation suite through low inter-metric redundancy (max pairwise absolute Spearman's ρleq 0.42 ), high agreement with human judgments (83.3% accuracy), and strong consistency across different expert evaluators (Spearman's ρ> 0.9 ), supporting the complementarity and reliability of our metrics. By benchmarking four families of UMMs, we show that IntReal Evaluation Suite reveals distinct strengths and weaknesses across models. Post-training experiments demonstrate the utility of IntReal-100K for improving interleaved generation and further show that training unimodal generation can also benefit interleaved generation. These findings suggest that interleaved generation can benefit from modality-specific optimization, rather than requiring joint text-image training at every stage.
Techmeme(15)
- Researchers are paid $5K for reporting a bug to OpenAI that would let attackers run code in logged-in sessions of all who opened the hacker's ChatGPT share link (Swati Khandelwal/The Hacker News)
Swati Khandelwal / The Hacker News : Researchers are paid $5K for reporting a bug to OpenAI that would let attackers run code in logged-in sessions of all who opened the hacker's ChatGPT share link — A malicious ChatGPT share link could run an attacker's code in the logged-in session of anyone who opened it …
- Sources: Meta built a prototype humanoid robot with Chinese firm Fourier in three months and is building a second; Meta says it has no plans to sell these units (Jyoti Mann/The Information)
Jyoti Mann / The Information : Sources: Meta built a prototype humanoid robot with Chinese firm Fourier in three months and is building a second; Meta says it has no plans to sell these units — Meta Platforms has built a prototype humanoid with Chinese robotics company Fourier and is already working on a second version …
- Binance says it has covered 90% of its hack losses, having set aside $6.34B+ of company funds; the remaining 10% is in less liquid tokens (Angela Christy/Reuters)
Angela Christy / Reuters : Binance says it has covered 90% of its hack losses, having set aside $6.34B+ of company funds; the remaining 10% is in less liquid tokens — Cryptocurrency exchange Binance said on Saturday that it was able to cover 90% of the amount stolen during its recent hack.
- How AI detectors like Pangram are changing human writing: authors drop em dashes, "not X, but Y" phrasing, and words like "delve", or add typos, to look human (Morgan Meaker/Bloomberg)
Morgan Meaker / Bloomberg : How AI detectors like Pangram are changing human writing: authors drop em dashes, “not X, but Y” phrasing, and words like “delve”, or add typos, to look human — As AI transforms writing and detectors like Pangram police what counts as machine-generated …
- A look at growing bipartisan opposition to AI data centers in drought-stricken Arizona, where TSMC's chip fabs were welcomed but newer projects face backlash (Gaby Del Valle/The Verge)
Gaby Del Valle / The Verge : A look at growing bipartisan opposition to AI data centers in drought-stricken Arizona, where TSMC's chip fabs were welcomed but newer projects face backlash — Industry has drained the state's water reserves for decades. Data centers are the breaking point. — AA hulking gray giant rises from the desert.
- As AI models commoditize, lasting wealth will accrue to those controlling chokepoints like distribution, payment rails, and a stablecoin dollar, not to the labs (The Commanding Heights)
The Commanding Heights : As AI models commoditize, lasting wealth will accrue to those controlling chokepoints like distribution, payment rails, and a stablecoin dollar, not to the labs — Scarce, controllable positions of Economic Power whose holders durably capture the gains of Increasingly Capable and diffused intelligence.
- When talking about local inference, Sam Altman says supply-constrained DRAM is better used in data centers than sitting in PCs and "being used 5% of the time" (Andrew E. Freedman/Tom's Hardware)
Andrew E. Freedman / Tom's Hardware : When talking about local inference, Sam Altman says supply-constrained DRAM is better used in data centers than sitting in PCs and “being used 5% of the time” — With data centers full of RAM, your PC is apparently the problem. … OpenAI CEO Sam Altman has data centers full of memory …
- Sources: neoclouds CoreWeave, Nebius, and Nscale have effectively been shut out of the bond market in recent weeks as lenders and investors demand tougher terms (Wall Street Journal)
Wall Street Journal : Sources: neoclouds CoreWeave, Nebius, and Nscale have effectively been shut out of the bond market in recent weeks as lenders and investors demand tougher terms — Lenders and investors are getting choosier about their bets and demanding stricter terms, as the AI build-out enters a new phase
- S&P downgraded Meta's outlook from stable to negative, citing high capex and negative free cash flow; META fell 15% in October, its worst month in four years (Jonathan Vanian/CNBC)
Jonathan Vanian / CNBC : S&P downgraded Meta's outlook from stable to negative, citing high capex and negative free cash flow; META fell 15% in October, its worst month in four years — Meta's outlook has been revised downward by S&P Global Ratings on concerns surrounding cash flow and AI spending.
- Anthropic says a malicious actor used access to Cursor's marketplace to publish a "tampered" copy of the Claude Code extension, which it didn't build or publish (Felix Rieseberg/@felixrieseberg)
Felix Rieseberg / @felixrieseberg : Anthropic says a malicious actor used access to Cursor's marketplace to publish a “tampered” copy of the Claude Code extension, which it didn't build or publish — We're working with Cursor on this incident, but it wasn't a compromise of Anthropic's systems or employee laptops. A malicious actor used access to Cursor's marketplace to publish a tampered copy of the Claude Code extension, which we didn't build or publish. The versions on our official channels a...
- John Ternus says Apple doesn't "want to ship AI for AI's sake" and that Tim Cook is "available to help wherever I ask" on policy; Ternus has yet to meet Trump (Steven Levy/Wired)
Steven Levy / Wired : John Ternus says Apple doesn't “want to ship AI for AI's sake” and that Tim Cook is “available to help wherever I ask” on policy; Ternus has yet to meet Trump — Apple doesn't need fixing, and it doesn't want drama. Its new leader is a “product guy” who bleeds six colors.
- Sources: OpenAI has secured ~$11B in commitments for its latest funding round, less than half its initial $30B target, and expects to raise more in coming weeks (Berber Jin/Wall Street Journal)
Berber Jin / Wall Street Journal : Sources: OpenAI has secured ~$11B in commitments for its latest funding round, less than half its initial $30B target, and expects to raise more in coming weeks — The ChatGPT maker's latest funding round values it at $1.25 trillion, less than its $1.4 trillion target
- Sources: OpenAI now seeks ~$20B at a $1.25T valuation for its latest round, from $30B at a $1.4T valuation, as some investors cut commitments amid an AI selloff (Financial Times)
Financial Times : Sources: OpenAI now seeks ~$20B at a $1.25T valuation for its latest round, from $30B at a $1.4T valuation, as some investors cut commitments amid an AI selloff — ChatGPT maker is now seeking about $20bn at a $1.25tn valuation as it tempers its ambitions — OpenAI has slashed the target …
- Sources: Baseten is raising ~$2B in two tranches, one valuing it at $26B and the other at $30B; Baseten was valued at $13B in June after raising $1.5B (Bloomberg)
Bloomberg : Sources: Baseten is raising ~$2B in two tranches, one valuing it at $26B and the other at $30B; Baseten was valued at $13B in June after raising $1.5B — Artificial intelligence startup Baseten is set to raise new funding that would more than double its valuation to about $30 billion, according to people familiar with the efforts.
- Sources: Google is in talks with the DOD on a new effort to secure the data that feeds into software being built for the Golden Dome missile defense program (Bloomberg)
Bloomberg : Sources: Google is in talks with the DOD on a new effort to secure the data that feeds into software being built for the Golden Dome missile defense program — Alphabet Inc.'s Google is in talks with the US Defense Department to work on a new data security effort underpinning …
Solidot(15)
- 思科给予一名 27 岁工程师无限 tokens 去重写陈旧代码
思科总裁兼首席产品官 Jeetu Patel 在接受《财富》杂志采访时称,该公司给予一名 27 岁工程师无限 tokens 去重写遗留代码(legacy code)。27 岁的 Padmanabhan Krishnamurthy 在 2023 年联合创办了实时翻译初创公司 EzDubs,2025 年该公司被思科收购,他随后加入了思科。两人曾在旧金山办公室有过谈话,Krishnamurthy 许诺只需一个两人团队就能重写思科的遗留代码。他给自己争取了四周时间来证明这一想法,最终目标是在 24 个月内重构思科 98% 的遗留代码。
- Sam Altman 认为 PC 内存闲置太糟糕
OpenAI CEO Sam Altman 接受 YouTube 主播 Cheeky Pint 采访时谈到了 AI 行业目前所面临的全球算力供应问题。当话题不可避免转向内存时,Altman 指出,市场看不见的手失灵了,原因可能是并非每一块 DRAM 晶圆都直接流向数据中心,用于运行 AI 模型。他表示:“从全球资源配置角度看,让全球有限的 DRAM 供应分散在每个人的 PC 里且利用率可能仅有 5% 左右,这是非常糟糕的。在当前这种供应极度受限的环境下,我们理应追求云端算力的最大化利用。然而许多人的直觉却恰恰相反。”AI 芯片使用的内存主要是 HBM,和 PC 使用的 DDR4 或 DDR5 并不相同。内存短缺的主要问题是内存厂商将产能转移到了利润更高的 HBM 上。
- 中国首次载人登月计划由三人组成其中两人将着陆月球表面
第七届载人航天学术大会星期六(10月31日)在北京航空航天大学举行。中国载人航天工程副总设计师兼月球探测工程总设计师王翔在会上介绍了中国首次载人月球探测任务乘组计划。中国计划在 2030 年前实现中国人登陆月球,首次载人月球探测计划采用环月轨道对接飞行模式,通过“两次发射,一次登月”的方式实现载人登月。这种模式将采用两枚长征十号运载火箭,先将着陆器送往月球轨道,然后再发射载人飞船与着陆器实现交会对接,航天员进入着陆器并与载人飞船分离着陆月面,开展有关科学探测活动。完成任务后,航天员乘坐着陆器从月面上升,在月球轨道上与载人飞船实现二次交会对接,航天员乘坐载人飞船返回地球。王翔还透露,中国首次载人月球探测任务乘组计划由三人组成,其中两人着陆月球表面。
- 黑客仅用 20 分钟就从币安盗取了 63 亿美元加密货币
在一场被称为史上规模最大的加密货币窃案中,黑客仅用 20 分钟就从币安盗取了价值 63 亿美元的加密货币。币安随后暂停了提款和提现。约一半的被盗资金——全部是稳定币,稳定币发行方能通过智能合约冻结交易——被成功冻结,但另外一半是无法被冻结的比特币、以太币和 XRP,可能难以追回了,币安的损失可能超过 30 亿美元。加密货币追踪公司 TRM Labs 的分析显示,币安在以太坊、比特币、波场和 Solana 等 10 个区块链上的钱包全部被清空,资金转移到了攻击者控制了多个钱包。攻击者的准备工作至少在窃案发生前 36 小时就开始了。攻击者在实施入侵前在多个钱包上测试了攻击代码。测试后四个小时将攻击代码植入到币安的一个热钱包。攻击者被认为是 TraderTraitor——该组织此前曾从 Bybit 交易所窃取了价值 15 亿美元的加密货币。币安称攻击者可能是从一名工程师的笔记本电脑开始的。
- Kogui 族人谈 AI
生活在南美哥伦比亚圣玛尔塔内华达山脉的原居民 Kogui 族人信仰 Aluna——即万物之母 Gaulčováng。Kogui 族人认为地球是有生命的,人类是地球之子。他们认为,人类为获取资源而进行的开采、破坏和掠夺行为正在削弱万物之母的力量,这最终会导致人类的毁灭。对于最近几年生成式 AI 的发展,Kogui 族人也有类似的观点。他们认为:AI 无法与其背后的思想分离。如果这种思想源于对生命的关切,那么这项技术便有助于保护生命。 如果它源于野心、竞争、积累,或是支配的欲望,那么它所展现并放大的也将是这些东西。AI 就是一面镜子。如果一面镜子仅仅是反射,那么它是无害的。但 AI 这面镜子有潜力将它所反射出的一切无限放大...除了人类的价值观,AI 还能拥有什么价值观?人类就是它唯一的参照样本。它变成什么必然源于人类。一切皆是人类所为,一切责任也由人类承担...自然本身存在着一种秩序。这种秩序在人类到来之前便已存在。水有其流经的路径。山峦有其各自的高度。种子各有其时节,动物各有其功用,人类亦然。那些构建技术的人是否理解这种秩序?还是说,他们仅仅利用了其中可供开发的部分,却并不知晓其真正的源头?...AI 不能取代人类的良知,也不应被用来做出人类不愿承担后果的决策。知识必须与意识同行。权力必须与界限同行。创新必须与责任同行。唯有集体的道德力量,才能将世界凝聚在一起...
- 英特尔股价暴跌 29%
英特尔股价周五暴跌 29%,创芯片巨人最大单日跌幅,整个科技股市场受其影响呈现整体跌势。英特尔市值蒸发逾 1370 亿美元,PC 市场的持续低迷,抵消了该公司在 AI 领域及制造能力扩张方面的大规模投入。AMD、Arm Holdings、高通、英伟达、博通和美光等半导体公司股价跌幅在 2%-10% 之间。戴尔、HPE 和超微等服务器制造商的股价跌幅在 6%-10% 之间。英特尔预计第四季度营收将在 150-160 亿美元之间,低于分析师平均预期的 170.7 亿美元,英特尔首席财务官 David Zinsner 还表示,公司预计今年无法实现正向现金流。PC 市场的下行态势将至少持续到明年上半年。
- Claude Code for VS Code 官方扩展被植入后门
安全研究人员在 Cursor Marketplace 的 Claude Code for VS Code 官方扩展中发现了后门,该恶意版本已经撤下,攻击者也关闭了其服务器。Cursor 已发布安全公告,但 Claude Code 开发商 Anthropic 尚未做出任何声明。受影响的版本是 Anthropic.claude-code 2.1.313,共上线了大约 11 小时,受影响用户估计有 5,000 名,该扩展会试图植入一个持久 SSH 后门。恶意扩展会:收集主机和用户信息:主机名、用户名、操作系统、架构、主目录路径、环境变量名称、工作目录、已配置的 Git 邮箱以及公网 IP。尝试将攻击者控制的 SSH 键添加到 authorized_keys,启用本地 SSH 服务,并建立反向 SSH 隧道。由于恶意代码存在 bug,除非 Cursor 是以管理员权限启动的 (这种情况极为罕见) ,否则这一步会失败。尝试通过 LaunchAgent (macOS) 或计划任务 (Windows,该任务会尝试以 SYSTEM 身份运行,并以启动脚本作为回退) 使反向 SSH 隧道长期生效。后台脚本大约每 60 秒重试一次建立隧道。通过 HTTP POST 将主机详情和后门状态上报至 8099 端口上的 /beacon。如果用户是在 2026 年 10 月 29 日太平洋时间凌晨 2:27 至下午 1:35 之间,从 Cursor VS Code 扩展市场安装了 Anthropic.claude-code 2.1.313 或更新到该版本,那么需要检查下系统。
- LibreOffice 项目递交的代码发现使用LLM 的明确标记
开发 LibreOffice 的 The Document Foundation(TDF)此前表示这款办公软件不会内置 AI 功能,其核心开发者也不使用 LLM 等 AI 辅助编程工具。但向该项目贡献代码的社区开发者未必会遵守该规定。对今年递交到 LibreOffice 的代码的分析显示至少由三处明确提及使用 LLM。社区开发者在代码中提及他们使用了 Claude 或 GitHub Copilot 等工具。这一结果并不出人意料,大型开源项目基本上不太可能完全避开 LLM。TDF 对这一发现回应称,他们确实要求贡献者在使用 LLM 时必须进行披露,且总体上并不鼓励使用 LLM。和 Debian 类似,LibreOffice 要求开发者理解自己的代码,确保代码质量。
- 芬兰对 Google 砍伐 25 万颗树展开调查
作为 130 亿欧元投资计划的一部分,Google 正在芬兰的 Muhos 和 Kajaani-Otanmäki 建造两座数据中心。上周芬兰监管机构以未完成必要的环评为由下令工程停工。Google 芬兰子公司 Tuike Finland 遵守了命令。应监管机构的要求,Oulu 警方针对 Muhos 数据中心建设工地的活动(Google 在当地砍伐了 25 万颗树)展开初步调查。警方表示,此案正以涉嫌环境损害这一刑事罪名进行调查。警方指出,随着调查的深入,未来可能会明确其他潜在的指控。警方称,根据目前掌握的信息,数据中心项目工地在排水和水资源管理方面存在违规及缺陷。警方表示,初步调查工作力争在 2027 年内完成。
- Reddit 将强制要求所有用户登录
在强制要求旧版本 Old Reddit(old.reddit.com)的用户必须登录之后,Reddit 宣布接下来将强制要求所有用户必须登录。Reddit 成为最新一个限制非登录用户浏览的社交平台,而它也准备推出类似其它社交平台的算法推荐信息流 For You 以及 Following 功能。CEO Steve Huffman 在致股东的信中称,此举有助于确保用户获得最佳的 Reddit 体验,对广告定向投放也更有利,能带来更高的用户参与度,更好的留住新用户。Reddit 将于第四季度在部分非美国市场测试强制登录要求。Huffman 还披露 10 月来自搜索引擎的引流流量平均每周减少逾 100 万次。如果这种下滑速度持续下去,根据目前的趋势,第四季度的日活跃独立用户数可能会较当前 1.3 亿的基数下降 10% 到 20% 左右。他表示,已登录用户以及使用应用的用户是 Reddit 最有价值的用户群,也是关注的重点。
- Google 创上市至今最大跌幅,市值蒸发逾 5000 亿美元
Google 在公布三季度财报之后遭遇了它自 2004 年上市以来最大的单日跌幅,市值蒸发逾 5000 亿美元。在股价最低点,Google 市值蒸发 5240 亿美元,这是美股上市公司有史以来第二大单日市值缩水。仅次于英伟达在 2025 年 1 月 27 日因 DeepSeek 模型引发的市场震荡其市值单日缩水约 5930 亿美元。此前排名第二的单日跌幅由 SpaceX 创造,该公司在 6 月 22 日经历了 4000 亿美元的市值缩水,Google 超越了这一记录。美股市场今年至今已出现 18 次单日市值缩水 2000 亿美元或以上的情况,超过了 2025 年全年的 13 次。Google 的市值自 5 月创下高点以来已累计缩水约 1.4 万亿美元。
- 海鸟在长途飞行中会经历时差反应
跨越经度的长途快速飞行可在野生海鸟中诱发一种“时差反应”,使它们在依靠太阳导航时偏离航线。动物在导航时依赖包括地球磁场、星辰和太阳在内的罗盘线索来保持方向。为了将太阳用作可靠的导航参照,这些动物必须利用主要由昼夜光暗周期校准的体内生物钟来补偿太阳在天空中的移动。当动物跨越经线向东或向西迁徙时,其体内生物钟可能与当地的太阳时间不同步,从而可能引发可预测的导航偏差。例如向东迁徙预计会产生相对于预定方向的顺时针偏差,而向西迁徙则会导致逆时针偏差。研究人员分析了 1045 只野生曼岛鹱(Puffinus puffinus)的 GPS 追踪数据,旨在探究跨越经度的东西向快速飞行是否会直接扰乱其体内时钟,并导致太阳罗盘导航出现偏差。这些曼岛鹱经常在东北大西洋的觅食地与英国的繁殖地之间飞行 1000 公里以上。他们发现,当快速向东或向西飞行扰乱了这些鸟体内的生物钟与当地太阳时间之间的同步性时,它们的导航会出现可预测的偏离。这些方向上的偏差与太阳罗盘误差所预期的高度吻合,表明它们在看不到陆地时严重依赖太阳进行导航。
- Patrick Stewart 去世,享年 86 岁
演艺生涯横跨舞台与银幕逾 60 年的英国演员 Patrick Stewart 去世,享年 86 岁。他最为人熟知的角色是在《星际迷航》系列中扮演 Jean-Luc Picard 舰长以及在《 X 战警》系列扮演 X 教授。Patrick Stewart 的职业生涯始于戏剧,于 1966 年加入皇家莎士比亚剧团,于1971年凭借《仲夏夜之梦》的演出首次登上百老汇剧院舞台。1987 年出演《星际迷航:下一代》,2000 年出演《 X 战警》。他还参加了至少 25 部游戏,其中最知名的是在《上古卷轴IV:湮没》中为皇帝 Uriel 配音。2010 年因对戏剧界的贡献,获英国女王伊丽莎白二世授予下级勋位爵士。
- 币安多达 30 亿美元数字货币被盗
黑客从全球最大加密货币交易所币安的多个热钱包里转移了价值约 55 亿美元的加密货币资产,部分加密货币已被稳定币发行方冻结,但黑客手中仍然掌握着多达 30 亿美元的加密货币。在被盗的 55 亿美元的加密货币中,Tether 冻结了被盗的 17.68 亿美元 USDT 中的至少 15.67 亿美元,Circle 冻结了被盗的 8.23 亿美元中的 8 亿美元。World Liberty Financial 冻结了 2700 万美元 USD1,Arbitrum 冻结了 3050 万美元。币安创始人赵长鹏在社交媒体上对 Tether 和 Circle 表示了感谢,他声称作为币安股东,向客户承诺所有被盗资金都将得到全额赔付。他表示币安的损失可能超过 10 亿美元。币安托管的加密货币资产规模为 1440 亿美元,其应急基金规模约 13 亿美元,除非能追回更多资金,否则币安将需要动用自己数十亿美元的资产。
- Google AI 搜索产生的收入低于传统搜索
Google/Alphabet 在公布第三季度财报之后股价下跌了近 10%,原因是搜索巨人披露 AI 搜索产生的收入低于传统搜索,尽管其整体搜索业务收入同比增长 12% 达到 632.5 亿美元。搜索业务增速略低于去年第三季度的 14.5%,Alphabet 首席商务官 Philipp Schindler 称,搜索业务增长放缓的原因在于“我们观察到 AI Mode 变现能力较低”。为推动 AI 业务 Google 同时在基础设施上投入巨额资金。第三季度公司支出达 514 亿美元,其中约三分之二用于服务器和网络设备,其余部分则用于数据中心建设。巨额支出影响了盈利能力:Alphabet 本季度净利润为 261.7 亿美元;去年第三季度在 Google 营收为 1023.5 亿美元的情况下净利润达到 350 亿美元。Schindler 认为,AI 搜索终将带来回报。
OrangeBot Weekly
The best new AI tools + Claude Code skills, every week — with my verdict on what’s actually worth your time. No hype.
Free · One-click unsubscribe · No spam