2026-10-28 的 AI 资讯

VCs will make billions on Anthropic’s IPO, but Amazon is set to make much (much) more

Massive note conversions reveal the true extent of Amazon's ownership in Anthropic.

Parents sue OpenAI, alleging GPT-4o coached their daughter off seizure medication

A family is suing OpenAI, saying ChatGPT coached their daughter to quietly taper off her seizure medication and keep it from her doctors before she died in February.

The data center boom has an upside: innovation for the US grid

As the AI boom stresses the existing power grid, all sorts of new opportunities for grid modernization are also emerging as a result, industry insiders say.

Qwen has alter egos triggered by flipping a single neuron

Hacker News 热榜讨论帖（21 分，8 条评论）

Expanding our partnership with GSA to bring Claude to all levels of government

Hacker News 热榜讨论帖（38 分，28 条评论）

Bank of America kindly asks that you don’t share your banking password with your AI assistant

The banking giant reminds customers that "you're responsible for activity" resulting from your AI agents' access.

Who’s really in charge at Anthropic after the IPO?

Here's a clear explanation of Anthropic's incredibly complex governance structure.

Early progress on accelerating drug development with Claude

Hacker News 热榜讨论帖（217 分，194 条评论）

Researchers are getting tired of OpenAI gatekeeping its breakthroughs

On Monday, OpenAI announced impressive breakthroughs in mathematics, computer science, and physics, along with what the company says is an evolving approach to communicating its findings. But despite its apparent course correction, researchers who spoke to The Verge are growing increasingly tired of the company's antics. It has been a turbulent few months between OpenAI […]

Raspberry Pi Zero 3W and Pi 4 Model A spotted in device trees

Hacker News 热榜讨论帖（155 分，80 条评论）

Survival LLM: AI assistant and emergency library that runs on your own computer

Hacker News 热榜讨论帖（325 分，136 条评论）

Disruptive raises $10B fund to make big bets on at least 10 companies

Disruptive has made headlines in recent years for pouring money into companies like chip startup Groq and the AI voice-generation startup ElevenLabs.

Honeycomb-Surfels: Compact Hexagonal Primitives for Efficient Scene Rendering and Reconstruction

Novel view synthesis and 3D reconstruction have advanced rapidly with radiance fields and splatting, yet their memory and computational demands limit their use in resource-constrained settings. We introduce Hexels: novel planar hexagonal surface elements with a first-order spatial color field for efficient 3D scene and 2D image representation. Hexels exploit locally planar regions with smoothly va

ThinkWorld: World Imagination Reinforces Chain-of-Dynamics Reasoning

Vision-Language-Action (VLA) models have made substantial progress in robotic manipulation, but their generalization remains constrained by the diversity and coverage of action-labeled robot demonstrations. Human videos provide much broader behavioral and environmental experience, while learning from them faces a fundamental trade-off between scalable data and explicit action grounding. We introdu

When Should We Intervene at Test Time?

Test-time adaptation helps a deployed model respond to changing inputs, but the same intervention can rescue one batch and damage the next. Updating normalization statistics, for example, can improve corrupted images yet fail when a batch contains mostly one class. Average accuracy hides these failures and cannot tell a model when to use an intervention. Target labels can guide this decision, but

Heads of Security: Learned Attention Gates Improve Prompt Injection Robustness

We introduce Heads of Security (HoS), a prompt injection defense that enforces instruction-data separation via a small set of learned per-head attention gates. HoS learns how each head should redistribute attention between trusted and untrusted tokens, enabling models to reason over untrusted content without following injected instructions. We evaluate HoS on Llama-3.3-70B-Instruct and GPT-OSS-20B

Attention as Lattice Gauge Theory: Relations Interact Only Through Wilson Loops

Relations compose through the entity they share - Alice is Bob's mother and Bob is Carol's father, so Alice is Carol's grandmother - and the order matters, since a father's mother is not a mother's father; verifying a composed relation means closing the loop. We show that if each entity's coordinate frame may be rotated independently (per-entity frame freedom), three things follow. In the pure-gau

KPPax: preprocess differentiable kinetic ODE for machine learning in atmospheric chemistry

Atmospheric chemical mechanisms are commonly distributed in the language of the Kinetic PreProcessor (KPP), which generates efficient Fortran or C code but does not fit into the differentiable programming frameworks in which machine-learning models are developed. We present KPPax, a kinetic preprocessor that generates differentiable, just-in-time compiled, and vectorisable chemistry code in JAX fr

Pushing Dexterity and Safety for In-Hand Object Pose Reaching That Works Remarkably Well

Dexterous manipulation with multi-fingered hands is the next frontier of robotics, yet learned hand skills remain limited to simple grasps or basic repeated motions, such as rotating an object. As robots attempt more demanding in-hand maneuvers, safety, such as not dropping the object, becomes critical, yet it is rarely measured. We study in-hand object pose reaching, a task that demands diverse a

GEPA-Curator: Curated Learning for Self Improving Agents

Large Language Models (LLMs) are increasingly used in multi-turn agentic workloads that require sustained interaction with complex environments. Existing reflective prompt optimizers such as GEPA learn from failed rollouts, but agentic failures can also come from prompt weaknesses, harness limitations, stochastic execution, or noisy judges. We argue that such failures may not provide useful learni

Can LLMs Uncover Hidden Physical Laws through Free Exploration?

Scientific discovery often begins with an unexpected observation that prompts a new research question. In a world with altered physical laws, can large language models (LLMs) notice unexpected behavior, such as heavier objects falling more slowly, and discover the law through experiments without being told what to investigate? To evaluate this ability, we introduce LongChang, an automated framewor

Scaling Laws for Tool-Use Agents via Controllable Environment Synthesis

We explore how to build performant large language model (LLM) tool-use agents that complete complex, multi-step tasks. Specifically, we study how data composition in synthetically generated tool-use environments affects the training of such agents. To do so, we propose the SynthAPI pipeline, which explicitly controls the diversity of generated environments along the axes of tools, tasks, and the d

MuonGS: Pretraining LLMs with Relaxed Weight-Norm Control

Weight decay and explicit weight-norm constraints improve optimizer performance in language model pretraining, yet a unified understanding of why such control helps and how much is needed remains incomplete. We analyze norm control in Muon over the entire training trajectory using the \emph{accumulated effective learning rate}, $C_T=\sum_{t<T}η_t/\|W_t\|_F$, where $η_t$ is the nominal learning rat

Multi-State Linear Attention with Generalized Second-Order Routing

Self-attention plays a key role in long-sequence modeling, but its quadratic time complexity fundamentally limits the scaling of Large Language Models. Recently, Multi-State Linear Attention (MSLA) mechanisms have been proposed to make sequence modeling more efficient by splitting the context into segments and compressing each in its own memory state. MSLA aggregates the information extracted from

Metacognitive Decoding: Efficient Inference for Diffusion Language Agents

Diffusion language models offer a promising path to efficient agents through parallel token generation. However, efficient interaction requires adapting both where computation is spent and how much text is generated. We introduce a metacognitive controller that guides diffusion generation according to the needs of the current response and the agent's interaction with the environment. It learns whe

What Improves During Pretraining? Tracking Knowledge Acquisition with Point-in-Time Models

Language models pretrained on web corpora collected over multiple years learn both general and domain-specific knowledge. How these different kinds of knowledge are learned as pretraining progresses remains unclear. Recent point-in-time models trained on web snapshots from different years provide a way to probe how knowledge prevalence varies over time. Using these models, we propose a temporal va

Rare, Relevant, and Still Wrong: How Memory Fails LLM Group Decisions

When an LLM group's discussion is compressed into a short memory for a final decision maker, what should the memory keep? Two natural heuristics keep the rarest claims or the claims relevant to the decision criteria. On hidden-profile tasks both look attractive, because how widely a fact is held is confounded with whether it helps. We introduce CrossProfile, a diagnostic benchmark of 60 templates

VideoLVR: Latent Visual Reasoning over Grounded Video Evidence

Chain-of-thought reasoning has substantially advanced video understanding in multimodal large language models (MLLMs). However, language-based reasoning traces do not provide a reliable mechanism for models to revisit relevant visual evidence, leading to visual forgetting and impairing their ability to reason over video content. Existing approaches maintain visual grounding during reasoning in two

EnvSched: Environment-Aware Scheduling for Rollout-Efficient Agent Reinforcement Learning

As executable environments for language agents scale, generating and verifying their rollouts becomes increasingly expensive. Under finite rollout budgets, this raises an important question: can data management at the environment level make agentic reinforcement learning (RL) more efficient by organizing, evaluating, and scheduling task instances together with the environments that determine their

Beyond Answer Changes: Matched Same-Class Controls for Local Audio Evidence

For audio-language models, an answer change after a local sound-event edit can reflect a response to the event's class, sensitivity to waveform rewriting, or both. We compare matched same-class and different-class replacements of a queried event. Both use the same interval, level estimate, donor fitting, and crossfade, while the same-class replacement preserves the correct answer. Across 600 synth

MoI-Bench: A Benchmark for Image Generation and Editing on Mobile Devices

Developing on-device image models requires considering practical user needs, task types and device resource constraints. However, existing benchmarks provide limited guidance on identifying capability gaps between on-device models and server-side models. We introduce MoI-Bench, a benchmark grounded in mobile user needs, comprising 64 subtasks and 1150 samples. It defines 16 operations, such as rem

GAS-PFN: Graph-Aware Spatiotemporal Prior-Data Fitted Networks for Zero-Shot Traffic Forecasting

Traffic forecasting requires modeling complex temporal dynamics and the spatial dependencies imposed by road topology. Supervised graph-based models achieve strong performance, but are often trained for a particular dataset and sensor graph, making deployment to new or evolving networks costly. Prior-Data Fitted Networks (PFNs) offer a zero-shot alternative by learning in-context inference from sy

VisionQuest is what happens when AI really goes rogue

In Marvel's VisionQuest, a superpowered robot tries to better understand humanity by getting lost in elaborate, digital daydreams in which he and other artificial intelligence-powered beings live together in a home. The series is deeply fantastical, yet it plays with some of the ideas that have come to dominate the conversation about real-world generative artificial […]

Understanding the limits of latent world models

World models are widely proposed as core components for enabling agents to reason over long temporal and spatial horizons. Yet current evaluations often emphasize task performance while providing limited insight into whether learned world models acquire predictive latent dynamics that reflect the underlying structure of the environment. Here we introduce the concept of faithful latent world models

Timestral

We introduce Timestral, a family of four time series foundation models for probabilistic forecasting, with 5M, 28M, 123M, and 209M parameters. Timestral sets a new state of the art on GIFT-Eval, TIME, and fev-bench, where Timestral-209M and Timestral-123M take the top two places on every metric. We achieve this with three ingredients: Decoupled Horizon Forecasting (DHF), which separates prediction

REGER: Modeling inter-item Relations with Semantic IDs for Generative Recommendation

Personalized recommendation depends on learning semantic and collaborative relations among items from online content and user interactions. Generative recommendation (GR) offers a promising paradigm by autoregressively generating discrete semantic IDs (SIDs) of items. However, randomly initialized SID token embeddings and GR models trained from scratch do not directly inherit the semantic relation

Zero-Shot Multimodal Event Spotting in Soccer Broadcasts

Automatic soccer event spotting supports highlight generation, video indexing, and sports analytics, but existing methods depend on large annotated datasets and soccer-specific training. We present a zero-shot framework that detects, classifies, and temporally localizes events in full-length soccer broadcasts without task-specific parameter training. Each match is segmented into overlapping clips

Spectral Braids: A Topological Descriptor for Dynamic Directed Networks

Eigenvalues are a workhorse of graph machine learning: spectral clustering, spectral graph neural networks, and graph kernels all summarize a network by the eigenvalues of a matrix. These eigenvalues are almost always treated as a static list of numbers. When a graph changes, however, its eigenvalues move, and when the matrix is not symmetric, they move in the complex plane, where they can wind ar

TeamGym: A Gym for Team Decision-Making by AI Agents with Distributed Information

Many real-world decisions require teams to integrate information distributed across participants. Yet AI agents are still primarily trained and benchmarked in isolation. We introduce TeamGym, an environment for evaluating and training language model-based agents on team decision-making problems constructed from real-world legal, finance, medical, and software engineering corpora. TeamGym introduce

LCM-Promo: An LLM-Based Uplift Model for Promotion Allocation

We present LCM-Promo, an LLM-based uplift model for promotion allocation. LCM-Promo is a task-specific Large Commerce Model (LCM): a finetuned LLM trained on transactional commerce data to predict user intent and promotion response. It learns monotonic promotion-response curves from production logs by interpreting single-token outputs as conversion-threshold probabilities, while conditioning on se

Gym-Anything-RT: Scaling Real-Time Environments for Vision-Language Agents

AI agents have made substantial progress in domains where they have a non-trivial amount of time to deliberate between actions, such as mathematics, software engineering, and performing tasks on desktop applications. It is less clear how such models fare in settings that demand real-time interaction, such as playing a video game or operating a physical device. In these settings, the environment st

Pneuma: A Large-Scale Real-World Time Series Benchmark for Commercial Aircraft Predictive Maintenance

Predictive maintenance is vital for commercial aviation, yet progress remains hindered by the lack of public, large-scale real-world datasets. As a result, most methods are developed and evaluated on proprietary data, limiting reproducibility. We introduce Pneuma, the first publicly available large-scale benchmark for predictive maintenance using onboard time series from in-service commercial airc

When uncertainty quantification fails: structural vulnerabilities and corrections for selective prediction under clinical class imbalance

Background. Selective prediction systems improve the safety of clinical artificial intelligence by deferring ambiguous cases to human experts while automating high-confidence decisions. However, this safety mechanism relies entirely on the premise that uncertainty scores accurately reflect clinical ambiguity. In prognostic tasks characterized by severe class imbalance, where the clinical outcome o

Frontier AI labs say their models refuted the Exponential Time Hypothesis. What does that mean?

TechCrunch spoke to the MIT computer scientist charged with explaining OpenAI's latest mind-bending discovery.

Anthropic’s elaborate plan to keep Wall Street at bay

Anthropic wants to have it both ways. With its impending IPO, it wants to tap the vast riches of the public markets and insists it doesn't want to compromise its core mission of developing AI safely and for the benefit of humanity. But greater model capabilities bring greater risks, and shareholders eager for returns may […]

Police are using AI to pull over drivers for 'suspicious' behavior

Hacker News 热榜讨论帖（370 分，283 条评论）

Google’s Genie 4 world model adds audio, more realistic worlds, and real-time steering

Genie 4 can maintain a consistent scene for up to two minutes, double Genie 3's limit, and offers better physics and more humanlike characters.

Amazon Is Gutting Its AI Division After Sustained Failure

Hacker News 热榜讨论帖（314 分，239 条评论）

Firefox’s Smart Window adds AI form filling and webpage monitoring

Mozilla is updating Smart Window - Firefox's AI-powered browsing experience - with a new feature that can help you fill out online forms based on your open tabs. As an example, Mozilla says Smart Window can pull information from your online portfolio or LinkedIn page to paste your work history into a job application. You […]

Genie 4: A new generation of world simulation

Genie 4: our new world model unlocking unprecedented realism in real-time simulation.

Google’s next Genie AI model generates more realistic experiences

Google announced today its Genie 4 world model for generating interactive experiences using AI. The company says the model has four main updates: It allows for more realistic worlds, lets users update an experience in real time with a natural language prompt, can generate audio, and uses Gemini 3.8 Flash under the hood to help […]

Anthropic trained Claude on illegally obtained Stanford professors' books

Hacker News 热榜讨论帖（520 分，274 条评论）

Retell uses OpenAI models to help power 100M monthly calls

GPT-6 Astra tops Retell’s voice agent benchmark, while Codex helps it reach a $1.6M annual revenue run rate per employee.

Meta AI director Rohit Patel leaves to take on Microsoft Office and Google Workspace

Rohit Patel is departing Meta after eight years working on AI evaluations and reinforcement learning, TechCrunch has exclusively learned. His startup, Ndoc, aims to take on Microsoft Office Suite.

Apple is adding an AI chatbot to Pages, Keynote, and Numbers

Apple is bringing an AI assistant to its suite of productivity apps, including Keynote, Pages, and Numbers. The assistant, which is built on AI models from Apple and OpenAI, can understand a document's context to perform tasks like designing, animating, and refining slides in Keynote. In Pages, Apple says the AI assistant can draft, restructure, […]

Japan internet giant GMO says hackers posted stolen customer data

Japan's GMO Internet is verifying alleged customer data posted by suspected hackers following a cloud service breach.

Here be monsters: What do you still need to know when agents do the work?

Hacker News 热榜讨论帖（169 分，93 条评论）

Opera introduces a new web standard to let users deploy AI locally in their browser

The new standard, WebAI, is designed to work with a variety of AI models.

General Compute raises $100 million Series A, countering the neocloud disaster narrative

General Compute is betting on inference chips from outside the Nvidia ecosystem.

Wikipedia's Cleanup of AI-Generated Content

Hacker News 热榜讨论帖（475 分，216 条评论）