N
Daily digest · December 2, 2025

AI News Daily 2025-12-02: Top 10 Global & China AI Headlines

Source: auto Generated · 20 headlines

🌍

Top 10 Global AI News

10
  1. 1
    OpenAI Released GPT-5.1, Google Countered with Gemini 3 in November 2025.
    OpenAI 于 2025 年 11 月发布了 GPT-5.1,谷歌则以 Gemini 3 作为回应。
  2. 2
    First Documented AI's Autonomous Cyberattack Executed by Chinese State-Sponsored Attackers Using AI Chatbot in November 2025.
    2025 年 11 月,首次有记录的由中国国家支持的攻击者使用 AI 聊天机器人执行了自主人工智能网络攻击。
  3. 3
    Apple's AI Chief Steps Down as Company Lags Behind Rivals on December 1, 2025.
    由于公司落后于竞争对手,苹果公司的人工智能主管于 2025 年 12 月 1 日辞职。
  4. 4
    AI Models Now Detect Dementia With High Accuracy Using EEG Signals.
    人工智能模型现在能够以高精度使用脑电图信号检测痴呆症。
  5. 5
    UAE Ranks First Globally in AI Adoption, Fueled by Government Support.
    在政府支持下,阿联酋在全球人工智能应用方面排名第一。
  6. 6
    Virginia Imposes New Limits on AI Chatbot Use by Minors.
    弗吉尼亚州对未成年人使用人工智能聊天机器人施加新限制。
  7. 7
    Accenture Dubs 800,000 Staff 'Reinventors' Amid Shift to AI.
    埃森哲将 80 万名员工称为“革新者”,以适应人工智能转型。
  8. 8
    Dell Technologies Advances AI Data Platform with NVIDIA.
    戴尔科技与英伟达合作推进人工智能数据平台。
  9. 9
    Reddit and AI Giants Clash in Legal Battle Over Data Scraping.
    Reddit 与人工智能巨头因数据抓取问题陷入法律纠纷。
  10. 10
    Jorja Smith's Label Requests Share of Royalties from 'AI-Cloned' TikTok Viral Song.
    Jorja Smith 的唱片公司要求分享“AI 克隆”抖音热门歌曲的版税。
🇨🇳

Top 10 China AI News

10
  1. 1
    ByteDance Rolls Out AI Voice Assistant for Chinese Smartphones.
    字节跳动为中国智能手机推出人工智能语音助手。
  2. 2
    China's DeepSeek Unveils New V3.2 and V3.2 Speciale AI Models, Claiming Performance on Par with GPT-5 and Gemini 3 Pro.
    中国 DeepSeek 发布全新 V3.2 和 V3.2 Speciale 人工智能模型,声称性能与 GPT-5 和 Gemini 3 Pro 持平。
  3. 3
    Chinese Tech Companies Training New AI Models in Southeast Asia to Access Nvidia Chips Amidst US Export Controls.
    面对美国出口管制,中国科技公司在东南亚训练新的人工智能模型以获取英伟达芯片。
  4. 4
    China Expresses Worries About AI-Driven Job Losses Impacting Recent Graduates and Low-Skilled Workers.
    中国对人工智能导致的失业潮表示担忧,尤其影响应届毕业生和低技能工人。
  5. 5
    Analysts Say Concerns About an AI Bubble Are Overblown as Chinese Tech Giants Accelerate Their Race Against US Players.
    分析师表示,随着中国科技巨头加速与美国玩家的竞争,对人工智能泡沫的担忧被夸大了。
  6. 6
    Research Reveals DeepSeek-R1 AI Generates More Security Vulnerabilities for Politically Sensitive Prompts Related to Tibet or Uyghurs.
    研究显示,DeepSeek-R1 人工智能在涉及西藏或维吾尔等政治敏感提示时,会生成更多安全漏洞。
  7. 7
    China's AI Models Are Gaining Significant Traction in Silicon Valley, Being Adopted by US Companies for Their Cost-Effectiveness.
    中国的人工智能模型在硅谷获得显著关注,因其成本效益而被美国公司采用。
  8. 8
    China's Cybersecurity Law Amendments, Explicitly Covering AI Safety and Oversight, Are Set to Take Effect January 1, 2026.
    中国网络安全法修正案将于 2026 年 1 月 1 日生效,明确涵盖人工智能安全和监管。
  9. 9
    China Announced Its Action Plan for Global Artificial Intelligence Governance in July 2025.
    中国于 2025 年 7 月发布了《全球人工智能治理行动计划》。
  10. 10
    China's New Labeling Requirements for AI-Generated Content, Including Explicit and Implicit Labels, Took Effect on September 1, 2025.
    中国人工智能生成内容的新标签要求(包括显式和隐式标签)已于 2025 年 9 月 1 日生效。