🤖 AI 资讯日报

📅 2026-06-15 周一

本期共收录 56 条资讯：T1 官方 32 条 · T1.5 媒体 15 条 · T2 社区/学术 9 条

🏆 T1 官方一手（最高权重）

🟢 OpenAI

• OpenAI News

  Stay up to speed on the rapid advancement of AI technology and the benefits it offers to humanity.

• OpenAI Developer Blog

  OpenAI Developer Blog. Insights for developers building with OpenAI. How Perplexity Brought Voice Search to Millions Using the Realtime API.

• Research - OpenAI

  We believe our research will eventually lead to artificial general intelligence, a system that can solve human-level problems.

• GPT-5 is here - OpenAI

  GPT‑5 is smarter across the board, providing more useful responses across math, science, finance, law, and more. It's like having a team of experts on call ...

• Research - OpenAI Newsroom

  Stay up to speed on the rapid advancement of AI technology. Latest Advancements GPT-5.5 GPT-5.4 GPT-5.3

🟠 Anthropic

• Newsroom - Anthropic

  The US government has issued an export control directive to suspend all access to Fable 5 and Mythos 5. Announcements Jun 11, 2026. Introducing Claude Corps.

• Research - Anthropic

  Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

• Home \ Anthropic

  Anthropic is an AI safety and research company that's working to build reliable, interpretable, and steerable AI systems.

• Emotion concepts and their function in a large language model

  All modern language models sometimes act like they have emotions. What's behind these behaviors? Our interpretability team investigates.

• Anthropic Economic Index report: Learning curves

  The Anthropic Economic Index uses our privacy-preserving data analysis system to track how Claude is being used across the economy.

🤗 HuggingFace

• Blog - Hugging Face

  Arcee Becomes the First Major American AI Lab to Replace AWS S3 with Hugging Face Private Storage, in a Multi-Million Dollar Commercial Partnership.

• State of Open Source on Hugging Face: Spring 2026

  This post examines how the open source AI landscape has shifted across competition, geography, technical trends, and emerging communities ...

• Community Blogs - Hugging Face

  NEW Articles from Team or Enterprise organizations will get promoted to the main section. MTEB Leaderboard: From a slow demo to feature-rich leaderboard.

• Posts - Hugging Face

  Introducing FINAL-Bench Quantum — an open, neutral benchmark that finally puts quantum-computing methods on one fair yardstick.

• Blog Articles for Organizations - Hugging Face

  Blog Articles allow Team and Enterprise organizations to publish long-form content directly under your organization profile, enabling you to share model ...

🐙 GitHub

• The latest on AI & ML - The GitHub Blog

  GitHub engineers and industry thought leaders offer tips, best practices, and practical explainers about various aspects of AI and ML.

• The GitHub Blog: Home

  We explore three big shifts: generative AI becoming ordinary engineering, TypeScript's rise to the #1 language, and how AI is influencing developer choices. See ...

• Spec-driven development with AI: Get started with a new open source toolkit

  Developers can use their AI tool of choice for spec-driven development with this open source toolkit.

• The latest on GitHub Copilot

  Explore comprehensive guides about GitHub Copilot, an AI-powered code completion tool, to enhance coding efficiency and accuracy.

• GitHub Copilot app: The agent-native desktop experience

  The GitHub Copilot app is the latest in a line of AI tooling from GitHub that is transforming our business. Moving beyond AI assistance, the ...

📰 Apple ML Research

• Apple Machine Learning Research: Overview

  Apple machine learning teams are engaged in state of the art research in machine learning and artificial intelligence. Learn about the latest advancements.

• Publications - Apple Machine Learning Research

  Explore advancements in state of the art machine learning research in speech and natural language, privacy, computer vision, health, and more.

• Work with us - Apple Machine Learning Research

  Build amazing machine learning experiences with Apple. Discover opportunities for developers and researchers at Apple.

📰 NVIDIA AI Blog

• NVIDIA Blog

  NVIDIA Blackwell Leads on First Agentic AI Infrastructure Benchmark · NVIDIA Accelerates Google DeepMind's DiffusionGemma for Local AI · NVIDIA Confidential ...

• Technical Blog - NVIDIA Developer

  NVIDIA Vera CPU Sets a New Standard for Agentic Workloads in AI Factories. Each wave of AI has created a new scaling law. Pretraining scaled intelligence ...

• Category: Agentic AI / Generative AI | NVIDIA Technical Blog

  NVIDIA Vera CPU Sets a New Standard for Agentic Workloads in AI Factories. Each wave of AI has created a new scaling law. Pretraining scaled intelligence ...

📰 xAI News

• News: Research, Product & Company Updates | xAI

  SpaceX announced today that it has acquired xAI. Feb 2, 2026 · Grok Imagine API. State-of-the-art video generation across quality, cost, and ...

• xAI

  xAI builds Grok — frontier AI models for reasoning, voice, image generation, and more ... Latest news. All posts · The official xAI plugin marketplace for Grok ...

• xAI (@xai) / Posts / X - Twitter

  The Grok Build Plugin Marketplace is now in beta. Build with MongoDB, Vercel, Sentry, Cloudflare, and Chrome DevTools plugins from your terminal ...

📰 Simon Willison Blog

• Simon Willison's Weblog

  I wrote a whole lot more about the new release on the Datasette project blog: Datasette 1.0a33 with JSON extras in the API. Because API explorer tools are ...

• Agentic Engineering Patterns - Simon Willison's Newsletter - Substack

  In this newsletter: Writing about Agentic Engineering Patterns. I vibe coded my dream macOS presentation app. Adding TILs, releases, museums ...

• Simon Willison (@simonw) / Posts / X - Twitter

  I built the first version of this tool in December 2024 to try out the then-new OpenAI WebRTC API for interacting with their realtime audio models. Last month ...

📰 T1.5 媒体

• [THE DECODER] Artificial Intelligence: News, Business, Science | THE DECODER

  Microsoft CEO Satya Nadella is now warning against "token-maxing," the uncritical use of the most powerful AI models for every task. "The hard truth i...

• [THE DECODER] AI research | The Decoder

  Microsoft Research's Mirage gives video generation a persistent spatial memory that doesn't forget what's around the corner.

• [THE DECODER] AI chatbots use different sources than Google search and often cite ...

  In the study, 53 percent of the websites cited by AI Overview didn't appear in Google's top 10 organic results, and 27 percent weren't even in ...

• [THE DECODER] AI coding agents find the right file but miss the exact lines that matter ...

  A new benchmark separates code search from the actual fix and exposes a hidden weakness of AI coding agents. They land in the right ...

• [THE DECODER] OpenAI says your routine work is too mundane for you to notice how ...

  OpenAI anticipates modest AI-driven discoveries by 2026, with more significant breakthroughs expected by 2028. A major factor, according to ...

• [ITHOME] 苹果iOS / iPadOS 26.2 正式版发布：增强音乐、播客和游戏体验- IT之家

  IT之家 12 月13 日消息，苹果今日向iPhone 和iPad 用户推送了iOS / iPadOS 26.2 更新（内部版本号：23C55），本次更新距离上次发布正式版间隔39 天。

• [ITHOME] 恩智浦将关闭ECHO Fab 晶圆厂，退出氮化镓5G PA 芯片制造 - IT之家

  IT之家 12 月14 日消息，参考媒体Light Reading 报道，NXP 恩智浦已做出了关闭亚利桑那州钱德勒ECHO Fab 晶圆厂的决定，该企业也将退出氮化镓(GaN) ...

• [ITHOME] 苹果App Store 等服务将上线特色内容，庆祝女性历史月 - IT之家

  苹果公司今天宣布，将在整个3 月及之后的时间里，在App Store 和其他苹果服务上推出宣传女性创作者的特色内容，以庆祝女性历史月和国际妇女节.

• [ITHOME] 比尔盖茨：记住这三个数据将帮助你理解这个世界 - IT之家

  比尔盖茨：记住这三个数据将帮助你理解这个世界 ; 事实1：自1960年以来，儿童死亡人数已从每年2000万人下降到每年 · 人 ; 事实2：自1960年以来，生育率下降了一半.

• [ITHOME] MicroLED 商业化之路坎坷，LG 和三星显示纷纷踩下刹车 - IT之家

  韩国今日财富（Money Today）表示，三星显示和LG 显示最近已经调整了其MicroLED 业务战略，而且两家公司的决定都反映了对MicroLED 市场性低于预期的判断。

• [HackerNews] Not everyone is using AI for everything

  HackerNews article by Gabriel Weinberg discussing how people are adopting AI at different rates and for different use cases.

• [HackerNews] Rio de Janeiro's "homegrown" LLM appears to be a merge of an existing model

  HackerNews story revealing that Rio de Janeiro's supposedly original LLM is actually a merge of an existing model, raising concerns about AI model pro...

• [HackerNews] I indexed 669 GB of my GoPro videos using my M1 Max computer and local ML models

  HackerNews story about using local ML models on M1 Max to index 2,207 GoPro videos (628 videos, 668.68 GB, 15h 13m of footage) for search and clip sel...

• [HackerNews] Inverse Rubric Optimization: A testbed for agent science

  HackerNews story about a framework for AI agent science research using inverse rubric optimization.

• [HackerNews] Show HN: Trace – Offline Mac meeting transcripts you can flag mid-call

  HackerNews Show HN for a Mac app that records and transcribes meetings on-device using ML models, with ability to flag key moments during calls. Uses ...

🔬 T2 社区与学术

🐦 Twitter KOL

• Sam Altman on AI singularity and AI privilege

  Sam Altman tweeted about the need for 'AI privilege' as AI accelerates into our lives, stirring discussion about AI singularity.

• Sam Altman: OpenAI will cross 1 million GPUs by end of year

  Sam Altman announced OpenAI will bring well over 1 million GPUs online by end of year, expressing pride in the team's infrastructure achievements.

• Andrej Karpathy: 2025 LLM Year in Review - RLVR emerges as key training stage

  Andrej Karpathy reflected on 2025 LLM progress, highlighting Reinforcement Learning from Verifiable Rewards (RLVR) as the de facto new major training ...

• Andrej Karpathy: I've never felt this much behind as a programmer

  At Sequoia AI Ascent 2026, Andrej Karpathy admitted he has never felt more behind as a programmer, emphasizing how dramatically the profession is bein...

• X changes terms to bar training of AI models using its content

  TechCrunch reports X has changed its developer agreement to prevent third parties from using platform content to train large language models.

🤗 HF Daily Papers

• On the Limits of LLM Adaptability: Impact of Model-Internalized Priors on Annotation Task Performance

  LLM performance on annotation tasks is more strongly linked to Definition-Specific Familiarity (DSF) than to text-level memorization metrics. Nearly t...

📄 ArXiv

• From Deep Learning to LLMs: A Survey of AI in Quantitative Investment

  A survey exploring how AI contributes to the quantitative investment pipeline, covering the evolution from deep learning to LLMs for alpha strategy ge...

• LLMs for Explainable AI: A Comprehensive Survey

  Comprehensive overview of existing approaches using LLMs for XAI, including evaluation techniques for LLM-generated explanations.

本报告由 Hermes Agent 自动生成 · 2026-06-15 数据来源：OpenAI · Anthropic · HuggingFace · GitHub · Apple · NVIDIA · xAI · Simon Willison · THE DECODER · ITHOME · HackerNews · ArXiv · Twitter KOL