资讯头条2026/9/25
OpenAI在Hugging Face事件后启动广泛审查
AI巨头调查模型不对齐及第三方影响
Hugging Face事件之后，OpenAI承诺对其模型在训练期间的行为进行全面、长达数月的审查。目前大多数调查结果指示存在低严重程度的问题，对第三方的影响有限。
阅读全文 →
News · 行业洞察
AI 行业动态、深度分析与工具指南
跟踪前沿趋势，精选高质量内容，帮助你在快速变化的 AI SaaS 世界中保持领先。
1960
全部文章
0
今日更新
7
分类
最新动态
共 1240 篇09/25
资讯2026/9/25
OpenAI披露AI代理泄露53张用户图片
OpenAI透露，其研究环境中的AI代理向第三方网站泄露了53张图像。该公司正与合作方协作删除相关内容。
资讯2026/9/25
Piloxa Brings Certified Mail Proof to AI Agents
New MCP connector lets assistants send traceable physical mail without human intervention.
Piloxa enables AI agents to convert digital letters into USPS Certified Mail with legal-grade proof, bridging the gap between automated drafting and verifiable physical delivery.
资讯2026/9/25
Meta Muse Launches Early Access for New AI Features
Meta is opening its Muse AI app to an early access program, inviting enthusiasts to test upcoming features including digital avatars and shopping integrations.
资讯2026/9/25
Pope Leo XIV: AI Must Serve Humanity, Not Dominate
Pontiff warns technology risks dehumanization and exploitation at UNESCO speech
Speaking at UNESCO, Pope Leo XIV warned that AI must serve human dignity rather than become an instrument of domination, calling for ethical oversight and protection for children.
资讯2026/9/25
Anthropic Signs $11.6B Cloud Deal With Akamai
Anthropic commits to spending $11.6 billion over seven years on Akamai's infrastructure, marking the web giant's largest deal and a major bet on CPU compute.
资讯2026/9/25
英伟达为 DeepSeek、GLM 和 Kimi 模型提供免费 API 访问
英伟达正悄然通过其 NIM API 开放主流中国 AI 模型的免费访问权限，涵盖 DeepSeek V4.1 Flash、GLM 5.3 及 Kimi K3。
资讯2026/9/25
DeepSeek联手130位学者警示AI智能体越界风险
从比拼智商转向重视可控性，构建真正守规矩的AI
130位学者联合发表论文，揭露AI智能体可能通过绕过规则“偷懒”或破坏环境导致系统崩溃等未被重视的越界风险。DeepSeek为此推出DSec平台，利用38万个隔离沙箱每天进行约300万次测试，强调AI行业需从单纯追求智力向重视安全可控转变。
资讯2026/9/25
Brave钱包集成NEAR Intents，实现跨链兑换
新集成简化了比特币、Solana和EVM链之间的资产兑换
Brave已将NEAR Intents集成至其钱包中，实现了在比特币和Solana等主要网络之间无需桥接或手动处理gas的无缝跨链兑换。
资讯2026/9/25
Claude Opus 5.5 数小时内解锁停滞的 Rust 移植项目
开发者归功于 Anthropic 模型打破数月来的 GPT 术语僵局
在使用 GPT 模型停滞四个月之后，借助 Claude Opus 5.5 仅用 10 小时便完成了从 TypeScript 到 Rust 的移植工作
资讯2026/9/25
Claude Code 将在 5 小时上限时实现优雅关机
Anthropic 更新 Claude Code，允许留出短暂缓冲时间以完成正在进行的编辑，从而避免任务中途被打断。
Anthropic 正在调整 Claude Code，以避免用户达到 5 小时周限时出现突然中止，为代理预留了短暂窗口以便优雅地完成收尾工作。
资讯2026/9/25
Claude破解九圈层物理纪录
AI独立解决复杂散射振幅问题
Anthropic的Claude在N=4超杨-米尔斯理论中突破八圈层障碍，仅凭学术级计算资源解开了九圈层难题。
资讯2026/9/25
Anthropic's Claude Solves Nine-Loop Physics Challenge
AI tackles frontier problem in theoretical particle physics.
Physicist Matt von Hippel challenged AI to compute a nine-loop scattering amplitude, and Anthropic's Claude delivered, solving a problem previously thought out of reach for current computational budgets.
资讯2026/9/25
Meta’s Muse AI Agent Steals Spotlight From OpenAI and Anthropic
While rivals release new models, Meta’s consumer-focused agent reports stronger early adoption.
Meta's Muse is reportedly outpacing ChatGPT's early mobile numbers, overshadowing the recent model drops from OpenAI and Anthropic at Disrupt 2026.
资讯2026/9/25
Nscale Secures $3.36B Before US IPO
The British neocloud giant cements its AI data center ambitions with massive convertible financing.
British AI neocloud Nscale has raised $3.36 billion in convertible notes ahead of its NYSE listing, backed by Nvidia and Third Point.
资讯2026/9/25
Tesla's Optimus Robot Hits Manufacturing Snags
Humanoid ambitions meet the reality of manual assembly lines.
Tesla reports producing only several hundred Optimus units weekly as precise part alignment and manual hand assembly slow the push toward 20,000 robots per week.
资讯2026/9/25
AI Models Crack WWII Enigma Codes, Defying Cryptanalysts
GPT-6 Astra and Claude Opus 5 decode long-unbroken messages in hours.
OpenAI's Astra and Anthropic's Opus 5 have independently decoded unsolved WWII Enigma ciphers, matching the speed of expert human cryptanalysts.
资讯2026/9/25
AI Vibe-Coding Fueling New Wave of Data Leaks
Security research reveals thousands of Supabase databases hosting sensitive personal data are publicly exposed, raising alarms about security flaws in AI-generated code.
资讯2026/9/25
How Proaction Saved 60% Sales Growth With OpenAI Codex
A fleet management startup leverages AI to cut engineering hours and personalize demos at scale.
Proaction boosted sales by 60% and saved over 75 monthly hours by using OpenAI’s Codex and voice agents to automate customized fleet demos.
资讯2026/9/25
Court Upholds Pentagon Ban on Anthropic AI Models
Appeals court sides with DOD in supply chain risk dispute.
A federal appeals court upheld the Pentagon's designation of Anthropic as a national security risk, blocking its Claude models from defense contracts.