AI 新闻 · V2

GPT Astra 会比你以为的更早发布。

OpenAI：我们终于造出了十个月前那篇经典博文《别造 Neuralese 模型》里警告过的 Neuralese 模型。

Fable 还行，但明天 Astra 会把它甩开。它轻松超出我对 AGI 的定义，而且幅度很大。 它有个性、有设计品味、超人级电脑操控与速度，当然在各个领域也都有大幅跃升。 发布已确认是明天，等待快结束了。 SSI 消息随后会有。

看起来 Astra 使用了 neuralese / 循环 Transformer，因此从根本上不再像以往模型那样拥有可完全监测的思维链。 这是非常非常坏的消息。 （链接：《信息报》关于 OpenAI Astra 秘密技术引发安全担忧的报道）

我想避免因混淆报道而引发一场冲向“不可监测性”的竞赛。我们目前前沿模型（包括 Astra）的计算图深度，仍在 GPT-4 的两倍以内。 自最早的推理模型起，OpenAI 一直致力于保留并利用思维链监测。我们非常重视这项技术，因为它能让我们看到模型对齐如何从训练分布泛化。我认为它确实脆弱，且不幸正朝负面方向发展，原因并不取决于架构变化——我很快会写。但我们仍有办法加强它，这也是我们当前研究计划的核心目标。

这是对可怕局面的精彩而精炼的表述。我唯一想补充的是，“高能力”系统按设计本就不该难以监控或对齐存疑。我们走错方向已经太远了。

很高兴你喜欢 3.7！

用相同提示词，由 Fable 5.1 生成的淹没新哥特塔楼 twigl shader。（可与引用推文中的 Fable 5 以及更早模型对比）

Grok 4.7 将在 10 天后发布

Dasseti提供面向投资顾问、机构投资者和资产管理公司的AI尽调及持续监控工具，交易条款未披露；目前命中主要为Nasdaq公告及其新闻稿转载，缺少独立媒体确认。

黄仁勋主张监管应聚焦已经出现的现实问题，而非假设性风险；Axios确认其出席同一场G20创新会议，但截至窗口结束，关于这段具体表态仅检索到Reuters稿件。

新原告包括学校教师、校长和学生，诉状指控OpenAI在内部人员建议报警后仍未联系执法部门；OpenAI否认部分有关管理层干预的说法。Global News在相近时间独立报道了新增诉讼。

交易包括约14.5亿美元交割现金及最多11.5亿美元业绩对价，目标是将微电网控制、现场发电协调和储能整合进Vertiv的数据中心电力方案；公司公告、SEC文件和Reuters报道一致。

报道分析了GPU小时价格指数和算力期货市场的发展，指出其可为数据中心贷款与长期算力租赁提供价格对冲，并可能进一步强化英伟达硬件的行业基准地位；Computable GPU Index及相关研究可验证该市场基础设施正在形成。

美国希望其他G20成员尽量沿用现有规则，仅针对AI带来的新问题制定监管措施；Reuters的会议报道确认了这一政策主张，Axios同时披露美国商务部与白宫科技政策团队存在路线和协调分歧。

OpenAI称Astra能自主发现并利用未知漏洞，最先进的网络安全能力首批仅向可信测试者开放；OpenAI官网披露与TechCrunch报道相互印证，但具体能力分数主要来自公司自测。

Fable 5.1面向普通客户和开发者，Mythos 5.1主要通过可信访问计划提供给网络安全及生命科学伙伴；缓存读取降价等调整可降低实际使用成本。Anthropic官网、Axios与TechCrunch均报道了此次发布及分级开放策略。

Training Physics-Informed Neural Networks (PINNs) requires jointly optimizing physics residual and initial/boundary condition loss terms, which often induce conflicting gradients. Gradient surgery methods mitigate this issue by constructing directions from loss-specific gradients to reduce conflict before optimizer transformation. However, even when the constructed direction is conflict-free, this

We present H3-World, an efficient framework that turns the 33B MiniMax-H3 video generator into an interactive world model. Our key finding is that, as large video generators become more capable, language is emerging as a natural interface for control. MiniMax-H3, for example, already supports zero-shot control of character behavior and camera motion through natural-language instructions. Building

Semiconductor supply chains face escalating risks from geopolitical tensions, geographic concentration, and rapid technological shifts, yet no scalable system continuously extracts, structures, and prioritizes risk intelligence from public corporate disclosures. We present an end-to-end pipeline that retrieves corporate documents for semiconductor companies and uses large language models (LLMs) to

Large language models (LLMs) struggle to classify text into taxonomies with many semantically similar labels, as the distinctions are domain-specific and not captured by pre-training. To handle large label spaces, a common approach retrieves top-$K$ candidate labels by embedding similarity and prompt the LLM to choose among them. However, top-$K$ retrieval reduces the number of candidates but does

Vision-Language Models (VLMs) provide useful priors for interactive decision-making, but using them directly as policies is expensive and brittle: they must be queried at every step, do not improve from environment interaction, and can repeat systematic errors. We study how to learn a cheap autonomous policy from an online, expensive, and imperfect but informative VLM teacher. We propose SAGE (Sel

Data-residency constraints force enterprises to self-host LLMs, but continuous adoption of newer models without decommissioning their predecessors expands the serving fleet, fragmenting a finite GPU pool. We consolidate traffic from over 200 internal applications onto a single model by closing quality gaps identified through production error analysis along three axes: instruction following, functi

How to divide a fixed annotation budget between supervised fine-tuning (SFT) and reinforcement learning (RL) during LLM post-training remains an open problem. Existing work characterizes only broad trends (e.g., SFT dominates in low-data regimes), lacks a principled allocation framework, and does not examine whether the optimal ratio transfers across model sizes. We frame this problem in terms of

Extracting structured fields from hundreds of millions of documents annually remains costly in regulated industries: bespoke OCR cascades cover only a fraction of workflows, privacy rules preclude external models, and existing open-source VLMs that clear quality thresholds cost more to serve than human annotation. We present a deployed document-understanding system built on a Mixture-of-Experts VL

Post-training quantization (PTQ) is widely used to reduce the cost of serving large language models (LLMs), but its accuracy cost is uneven and is often tuned per model. We study where quantization damage occurs and how to allocate a small additional precision budget. Using causal mixed-precision intervention as ground truth (raise each layer to 8-bit in turn and measure the accuracy it recovers)

Writing involves diverse cognitive activities, from ideation to revision, and writers' needs vary across individuals and moments. Proactive AI promises to provide the right support at the right time, yet existing proactive tools largely focus on generic textual assistance, such as autocomplete. This paper studies the design space of proactive thought partners: AI agents that proactively offer cust

AI tutors are most useful when they adapt to each student's strengths, weaknesses, and preferred guidance, but evidence about which guidance works for which student is sparse, slow, and costly to collect from real learners. Student simulators can provide this signal as a proxy, yet existing approaches are limited: state-tracking models fit student behavior but struggle to process explanations or c

We develop a framework for mechanism design with AI agents whose alignment (preferences) and capabilities (feasible actions and information) are unknown. We want such agents to act on our behalf so mechanisms must incentivize both honesty and obedience. A one-sided imitation structure---capabilities can be concealed but not counterfeited---yields a revelation principle, a characterization of imple

Real-world robotic assembly at sub-millimeter tolerances demands spatial precision, compliant interaction, and robustness to contact failures. We present Facet-0, a robotic foundation model that predicts and values the contact consequences of its actions. Facet-0 unifies multimodal representation learning and reinforcement learning (RL) post-training around a joint action-wrench proposal: a causal

Natural language is emerging as a primary feedback channel for improving language agents, capable of conveying intent, preferences, and causal structure in forms interpretable by both humans and modern language models. We call this paradigm Verbal Reinforcement Learning (VRL) and offer the first unified account of it. We organize the field around a single axis, \textit{when} verbal feedback takes

Dynamic agent harnesses let language models change the software that shapes their own execution. This flexibility brings a new reasoning burden: a local plugin change can propagate through dependencies and cleanup. We introduce CordisBench, a 1,200-question benchmark of this lifecycle reasoning. It combines a controlled formal setting with programs executed against Cordis, a runtime that manages c

The repository-level code generation task requires synthesizing code that satisfies task requirements while remaining consistent with the target repository context. Since real-world repositories often exceed the input length limits of LLMs, existing approaches commonly adopt retrieval-augmented generation (RAG) to provide repository-specific context. Despite improving repository-context retrieval,

Evaluating software engineering agents on realistic benchmarks is costly, since each task may require multi-step code exploration, modification, and test execution. Existing efficient evaluation methods select representative subsets to estimate full-benchmark performance, but are largely result-only: they fit historical pass/fail response matrices or static task semantics, discarding how agents so

LLM-based evaluators of natural language generation (NLG) quality are widely deployed as scoring tools and as automated training signals, yet the internal procedure by which they assign a rating remains poorly understood. We investigate this procedure mechanistically through an eight-attack perturbation taxonomy across the Readability and Adequacy dimensions of NLG quality, a generation pipeline t

恰逢其时，谷歌据称即将发布以编程为先的AI模型 - Gizmodo

芝加哥大学推出应用人工智能新硕士项目 | Newswise - Newswise

评论 | 对人工智能的人为批评？ - WSJ

洛杉矶联合学区课堂将禁止人工智能吗？新委员会将审议相关规定 - ABC7 Los Angeles

自由职业者被“毫无灵魂”的AI垃圾内容清理工作淹没：“很遗憾我们不得不做这些” - The Guardian

Eco Wave Power与AI engineering GmbH签署协议，推进波浪能AI数字孪生平台 - Yahoo! Finance Canada

报告：自2024年以来AI岗位薪资飙升23%，传统科技岗位薪酬下滑 - ET CIO

物理AI与众不同，正在改变国防系统的设计方式 - Breaking Defense