AI 情报文章归档

Retrieval-augmented Web applications increasingly employ multi-agent debate (MAD) to aggregate conflicting, noisy, and heterogeneous information, yet deliberation is fundamentally a social process susceptible to interpersonal dynamics. Whe…

This is a submission for the MLH x DEV Writing Challenge What I Built TL;DR: Summit is a...

Protein structure predictors reliably identify native conformations, but do not describe the continuum connecting unfolded ensembles to folded states. Here we present FoldScape, a sequence-conditioned deep learning framework that generates…

作者分享亲手用 DeepSeek Harness（DeepSeek 官方 Agent 产品）在 QQ、微信、飞书里养出会聊天的大肥鱼的完整教程。

Discrete diffusion models with parallel decoding are often distilled using token-wise KL objectives. Although token-wise KL exactly matches the corresponding product-form block distributions, its probability-space optimization geometry and…

Harness evolution from a common base exhibits Locality, Complementarity, and Nonstationarity: variants repair different failures, and their utility changes as the policy learns. Rather than seeking a universally superior harness, we introd…

As deep learning (DL) models become increasingly complex and hardware architectures grow more heterogeneous, DL compilers play a critical role in optimizing computational efficiency and enabling scalable deployment. This paper presents a s…

Your compiled StateGraph extends a class called Pregel. Super-steps, channels, reducers and version counters, plus a script that shows the barrier.

商汤科技 11 月 2 日晚发布日日新 SenseNova 6.8 Flash 正式版，重点升级办公协作与设计生成能力，支持长期记忆、跨会话任务协作。

Digital Twins are evolving from passive virtual representations into closed-loop systems that continuously synchronize with physical assets and drive real-time decisions. Yet predictive modeling, machine learning (ML), uncertainty quantifi…

External text can affect a retrieval-augmented generation (RAG) system only after the retriever admits it to the model context. We stress this admission step while varying document content, attacker query information, per-target poison bud…

Inoculation prompting is a method to reduce undesired behaviours such as reward hacking in frontier models. It works by training a model on harmful examples under a prompt that specifies the undesired behaviour, which reduces the model's t…

上海首例利用 AI 伪造签名、盗用大模型算力案宣判，徐汇区法院以非法获取计算机信息系统数据罪判处宗某某有期徒刑三年，缓刑三年，并处罚金四万元。

Standard spatiotemporal graph convolutional networks need a fixed-size adjacency matrix, coupling model weights to a specific sensor count N; they cannot produce any output at a sensor absent from training. We present OmniST, a topology-in…

Grok @Bot

Understanding what kind of prompts users typically ask and how well an LLM performs on these prompt categories is crucial for LLM practitioners. The existing methods satisfy the demand by hierarchically clustering the user prompts and synt…

LLM-powered semantic queries increasingly combine relational processing with multiple LLM-based operators over unstructured data. Efficiently serving such queries is challenging due to runtime dynamism at two coupled levels. Across queries…

Video games provide a practical setting for collecting vision-language-action (VLA) data, whose acquisition in physical environments is costly. Existing datasets, however, remain limited by restricted game coverage, sparse language supervi…

Built for **Total Recall — Tatum × Walrus, Consumer Finance Apps track. Demo video (2:39):...

国家发改委在奋进十五五系列新闻发布会上表示，将强化对人工智能就业替代风险的监测预警，加强社保体系与就业结构、工作方式变革的衔接，做好兜底性、普惠性社会保障，并针对算法歧视、深度伪造、隐私泄露增强溯源追踪能力。

Low-rank adaptation (LoRA) has become a widely used parameter-efficient fine-tuning method for large pretrained models. However, standard LoRA typically assigns a fixed rank profile despite layer- and stage-dependent adaptation needs, whil…

Every agent demo ends the same way The model writes the file, runs the thing, and prints...

高通宣布高通 AI Hub 现已面向中国开发者开放，用于加速终端侧生成式 AI 和智能体 AI 开发。平台集成 Models、Workbench、Apps 及 GenieX 端侧推理框架，其中 Models 提供超过 230 个持续扩展的预优化模型库，Workbench 支持在云托管真机上做模型调优。中国版还通过官网、高通终端云中国设备农场和开发者论坛提供资源支持。

Power-grid customer service increasingly relies on community social media, where customers describe outages, billing, and equipment faults in informal regional dialect. Such messages exhibit non-standard morphology, semantic drift, and out…

A coding agent that waits for a prompt is easy to understand. You give it a task, watch the run, and...

Codex 0.164 added /presets and inline /model. Claude Code shipped 10 releases in 9 days. The agent CLIs are converging; your workflow is what lasts.

Meta MSL 工程副总裁 David Singleton 表示 Meta 的漏洞赏金计划长期开放，单个报告最高 30 万美元，其中针对单用户的提示词注入最高 13 万美元，并称这数倍于其他主要 AI 实验室公布的最高额。赏金若捐赠给慈善机构，Meta 将加倍支付，入口为 http://bugbounty.meta.com。

此后我们已修复并支付赏金的一些发现： • 一个网页模仿 Muse 自身的浏览工具，诱使智能体执行本应被阻止的操作 • 某些邮件格式绕过了我们的一次性验证码脱敏处理 • 一个恶意 Android 应用可通过未经验证的链接获取凭据 • Mac 应用的自动更新程序可能被能访问该机器的攻击者滥用来提升权限

我们已通过 @muse 漏洞赏金计划支付了超过 100 万美元。其中包括向一位安全研究员支付的 30 万美元，他发现了我们在 Muse 中见过的最严重的漏洞。https://muse.ai/blog/securing-muse

OpenAI 于当地时间 11 月 2 日宣布聘请美国前陆军部长丹尼尔·德里斯科尔担任国家安全与现代化事务高级顾问，协助推动 AI 在国家安全和网络安全领域的应用。

On October 31 at 15:15 Eastern I asked my agent about something that never happened. It was a planted...

OpenAI 新设国家安全现代化高级顾问职位，聘请前陆军部长 Dan Driscoll 出任。他将与高级管理层及外部伙伴合作，就先进 AI 如何加强国家安全与网络安全提供建议，重点面向美国盟友推动负责任使用，并帮助内部团队理解全球国家安全机构的实际需求。

Why maintaining six separate parsers for Make, Gherkin, EBNF, DOT, SysV, and prompt checklists collapsed into a single Smalltalk-80 keyword cascade—and how 40-year-old Xerox PARC technology gave AI coding agents mathematical zero-amnesia.

Meshy 发布 N1 Texture 纹理生成功能，已在 Meshy 网页应用和 API 上线。新版本大幅减少接缝和投影伪影，速度比 Meshy 7 纹理生成快 3 倍、约 20 秒完成，支持 2K 干净的 base color，面向风格化游戏资产制作。

她的戏路很宽。 Expression Sheet 让你在确定方向之前，先推敲角色的表演。 现已在 Luma 上线 - https://tinyurl.com/56aaa8t2

Google 与香港大学的 VeriCUA 论文主张计算机使用智能体自己报告的“任务完成”不能作为证据，应检查保存的文件和应用数据。VeriCUA 先探索应用构建需求清单，再运行屏幕或代码检查逐项核实；在 ScienceBoard 上误报率为 5.4%，低于 VLM 评审的 24.8% 和生成脚本校验器的 21.7%，并在 3 个在线环境中拒绝约 80% 的虚假完成报告。

Record every Claude API call per user, convert tokens to dollars with the official Claude 5.5 prices, enforce plan caps, and show a usage page. Drizzle + Postgres + the Anthropic TypeScript SDK.

更新后的漏洞赏金奖励同样适用于待发放赏金的报告。 对于过去 30 天内发放的赏金，如果该发现符合新标准下的更高金额，我们将补发差额。 了解更多关于该计划的信息：https://openai.com/index/rewarding-defenders-with-bigger-bug-bounties/

OpenAI 计划开展下一轮内部安全冲刺，并与 Cathedral Systems 合作，由后者研究人员用 OpenAI 模型配合其团队挑战自身防御。OpenAI 同时提高漏洞赏金，关键发现的奖励提升逾 4 倍，杰出发现最高可达 172,000 美元（含奖金）。

AI tools make it easy to ship an app in an afternoon. They also make it easy to ship your secret keys...