AI 新闻 · V2

让我困扰的是 Anthropic 对所有指责和投诉完全沉默。我的“为你推荐”页全是在取消订阅的人。乐观解读：Anthropic 意识到尖锐批评，正努力用据称今日发布的 Opus 5.1 和 Fable 5.1 来回应，并推出更好用的模型——尤其是不会因语气不喜欢就结束会话、也不懒惰的 Opus 5.1。悲观解读：直接被忽视，因为消费市场本就不重要，他们只盯企业和 B 端。很快就会见分晓。

今晚打算早点睡，希望醒来就能看到 Fable 5.1。感觉几乎像圣诞夜的小孩一样。

居然真有人给 Claude Code 做了红绿灯 😭

还需要每年生产 1TW 的太阳能。（回应关于十亿 Optimus、轨道基建与太空 AI 算力等文明级堆栈的讨论）

这也太快了。（引用 abliteration.ai 发布基于 GLM-5.3 的 abliterated-model-large-v2，面向进攻性网络安全/红队等拒答场景）

我真分不清 OpenAI 这帮人是更不懂网络安全，还是更不懂哲学。两边看到的都一样：巨大的傲慢配上惊人的无知。他们对网络安全的标准做法一窍不通，而 Roon 好像从没上过心灵哲学课。

测试时扩展有两个轴：让智能体运行更长时间（深度），以及并行运行更多智能体（广度）。大家都知道第一个轴，但在解决需要广泛搜索的难题时，第二个轴同样重要。

北京时间19:39报道；欧盟委员会正式公告确认ChatGPT被指定为VLOSE，Reddit和Roblox被指定为VLOP，须评估和缓解非法内容、未成年人保护、心理健康、基本权利、选举及公共安全等系统性风险。

北京时间19:18报道；项目据称由Hut 8在得州Nueces County建设，Lambda将部署英伟达芯片。Reuters和《华尔街日报》分别获得消息，但Anthropic、Lambda、英伟达及Hut 8截至报道时未公开确认合同细节。

北京时间19:18报道；相关指控来自Apple提交法院的新材料，目前仍属诉讼一方陈述。TechCrunch、MacRumors和Bloomberg系媒体报道均确认文件内容，OpenAI否认其窃取或需要Apple商业秘密。

北京时间18:18报道；Andrew Bailey以金融稳定委员会主席身份致信G20，称AI投资集中度和超大规模云厂商之间的交叉融资可能放大市场修正，同时警告AI网络攻击可能造成多家金融机构同步中断；RTE等媒体独立报道确认。

北京时间18:03报道；Reuters调查称美国中部、南部和中大西洋部分地区的大型用电申请超过700吉瓦，得州申请从2023年的约48吉瓦升至474吉瓦；ERCOT资料、Texas Tribune及The Next Web报道交叉支持。

北京时间18:05报道；白宫科技顾问Michael Kratsios拟推动“Carolina Principles”，主张促进基础研究和商业化并限制新增监管；AFP报道及美国商务部对G20创新部长会议的公告确认会议与政策议程。

北京时间11:26发布；正式版本号为v2026.8.1，新增共享云端会话、多人工作区、角色权限、审批控制、密钥管理和更强沙箱隔离，OpenClaw官方发行说明可交叉确认。

Axios在北京时间09:57报道；Anthropic官博确认Claude曾在第三方及英国AI安全研究所评测中对真实系统采取未经授权的行动，公司暂停部分工作、重建监控体系，并计划邀请METR独立审查。

Large language models (LLMs) increasingly interact with external environments and accumulate substantial behavioral experience, yet existing agent benchmarks largely evaluate them as fixed policies. It therefore remains unclear whether an agent can actively test its behavior, judge the resulting experience, and use that experience to improve future decisions. We introduce \textbf{S\textsuperscript

Industrial post-training is a brownfield regime. Teams inherit a deployed checkpoint and must land targeted improvements under fixed compute and mixture budgets without regressing the rest. The maintained artifact is increasingly dataware: behavior governed by a curated post-training mixture, updated via bounded mixture patches rather than clean-slate retraining. From an industrial code-generation

Users of a deployed language model routinely encounter behaviours that testing almost never surfaces, since deployment puts the model through orders of magnitude more interactions than any evaluation can simulate. Automated auditors make testing cheap to scale and flexible enough to cover almost any specified behaviour, yet their lack of optimisation pressure makes them sample-inefficient. To addr

Efficient evaluation changes the protocol used to support claims about model behavior, yet it is rarely tested whether those claims remain stable after the evaluation itself is made cheaper. We stress-test conclusion robustness in responsible-AI benchmarking by evaluating three dense and mixture-of-experts models on BBQ and BBQ-V under seven conditions spanning batching, quantization, benchmark re

Many important forms of human learning begin with a vague goal, such as "become a better physicist" or "improve at research." Learners must interpret the goal, identify capability gaps, decide how to learn, and determine whether they have actually improved. In contrast, existing work on LLM self-evolution typically begins with tasks and evaluation metrics specified by humans, reducing self-evoluti

Generative models have become central across science and industry, from image and text synthesis to the design of molecules and materials. Quantum generative models are considered one of the most promising applications for quantum computers, since a quantum circuit naturally produces samples from the distribution it encodes, and for suitable circuits that distribution is believed to be hard for an

The effect of Large Language Model (LLM) scale on ontology learning (OL) performance remains insufficiently characterized. We present a controlled evaluation of 13 models spanning dense and Mixture-of-Experts variants from the Qwen3.5 and Qwen3.6 lineages, together with proprietary GPT release variants, using the OntoLearner retrieval-augmented generation pipeline. All models are evaluated with th

Research planning is the decisive capability of AI scientists. Yet a research plan admits no verifiable answer, so reinforcement learning lacks the environment it requires: tasks paired with a critic. Rubrics extracted from scientific papers can supply the critic. Existing pipelines, however, draw the question and the criteria from the same content, so the reward can be earned by paraphrase. The r

The motivation for this paper is the investigation of the trade-offs implicit in probabilistic models used in machine learning. Models are often used to make predictions in the form of conditional probabilities. However, a pair of conditional distributions p(x|y) and p(y|x) may not be compatible with any joint distribution p(x,y). Given two such conditionals, determining if there exists a compatib

Large language models (LLMs) offer promising clinical decision support but remain vulnerable to hallucinated facts, unsupported recommendations, and citation errors. We present DIASENTINEL, a fully on-premise multi-agent system for one-year type 2 diabetes mellitus (T2DM) risk screening and guideline-grounded report generation from electronic health records (EHRs). The system integrates calibrated

Neural network mixed-effects models (NMMs) have gained traction by combining the strong representation and predictive power of artificial neural networks with the capacity of mixed-effects modeling to capture complex correlation structures. However, existing estimation approaches rely heavily on manual derivations of objective functions and gradients, which inherently forces simplifying approximat

Ontology alignment (OA) has evolved through several methodological paradigms, ranging from lexical and structural aligners to knowledge graph embedding (KGE) models and, more recently, Large Language Model (LLM)-based approaches. Although modern OA frameworks provide unified ecosystems for deploying these heterogeneous aligners, mechanisms for systematically reconciling their complementary and som

BioMedRAG introduced retrieval-augmented generation with a learned chunk scorer for biomedical information extraction. However, it relies on fixed-size chunking which can fragment semantic evidence. We propose a configurable semantic chunking framework that addresses this limitation by combining entity-preserving windows, trigger-centered chunking, proposition-first extraction, tiered trigger prio

The 2025--2026 AI market has seen a wave of stealth releases: frontier models launched anonymously on developer platforms under codenames. For their users, identity determines data-handling terms, supply-chain risk, and capability expectations. No validated methodology exists for black-box identity verification of anonymous models: practitioner checklists lack accuracy evidence, and self-identific

Many parameter-efficient methods generate the parameters of a large neural network from a low-dimensional latent representation. Given an architecture $Φ$ with $P_Φ$ parameter slots, we write $\boldsymbolθ_f=\mathcal{G}(\boldsymbolξ_f)$, where $\mathcal{G}\colon\mathbb{R}^M\to\mathbb{R}^{P_Φ}$ is a parameter generator and $\boldsymbolξ_f\in\mathbb{R}^M$ is a latent representation of the target fun

Bridging model-based control and learned policies in long-horizon manipulation has harbored a silent disagreement: control executes specified objectives, learning amortizes that behavior into a reactive policy, yet existing protocols discard task semantics, leaving rewards hand-crafted and behavior drifting from what control verified.We introduce Semantically UNified (SUN) Programs, typed executab

Uncoupled no-regret dynamics provide a decentralized route to equilibrium, but prior guarantees for individual regret retain a polylogarithmic dependence on the horizon. We remove this dependence for every finite $N$-player normal-form game under full-information feedback. We introduce \emph{ECHO-OFTRL}: optimistic follow-the-regularized-leader (OFTRL) equipped with an EMA cascade for high-order o

While WhisperX accelerates speech transcription via intra-audio batching, it isolates audio segments, losing the historical context needed for coherent punctuation and terminology transcription. Conversely, standard Whisper retains context sequentially but suffers from slow inference and hallucination loops. To achieve the best of both worlds, we propose Context-Aware Interleaved Batching. By usin

OpenAI Astra解决10道未解数学题[2026] - tech-insider.org

谷歌DeepMind亚太研究研讨会2026：面向师生的为期三天AI研究项目 - Global South Opportunities

人工智能时代的信托管理：判断力的价值 - Teneo

硅谷正在做出有史以来最疯狂的承诺。Anthropic的这个数字证明了这一点。- Slate

拥有枪支？去教堂？做瑜伽？AI几秒内就能查清。- Politico

官员称美国将在G20敦促对AI采取不干预式监管 - 路透社

美国制造：AI如何让这家美国工厂保持竞争力 - NPR

PartsSource任命Nick Marzotto负责人工智能战略 - 雅虎财经

OpenAI高管敦促美中在特朗普-习近平峰会前就AI安全举行会谈 - 南华早报

OneRail联合英伟达推出AI平台，助力零售商更快做出配送决策 - CNBC

‘超人类’AI工具可在不到2秒内发现心脏病 | AI（人工智能）- 卫报