← AM·PM Brief

Anthropic Blames 3 China LLMs of Distill Theft; UK Cops Bag £800K Scam with Palantir AI

· Evening brief · 10 news · 2:53

Audio in Mandarin Chinese · English transcript below

US-China LLMs clash: Gemini3.1 vs GLM-5 vs Qwen3.5; AI builds SaaS, solves crimes, delivers food, deletes your inbox in 1h!

In the AI sector today, the boundaries of technological competition and the deepening of practical applications have become central focal points. Anthropic's accusation that Chinese models like DeepSeek engaged in model distillation not only thrusts AI intellectual property into the spotlight but also sparks significant industry concern over compliant development. Concurrently, UK police, leveraging Palantir AI, successfully cracked an £800,000 fraud case, powerfully demonstrating AI's immense potential in social governance and crime fighting. These two developments, juxtaposed, clearly delineate that as AI rapidly evolves, it faces challenges related to ethical norms while also showcasing broad prospects for societal empowerment.

Today's Top 3 Headlines

  1. AI争议

    Anthropic(AI公司)指控DeepSeek、Moonshot、MiniMax模型蒸馏

    Anthropic(AI公司)指控中国AI公司DeepSeek、Moonshot和MiniMax非法窃取其Claude(AI助手)模型能力,引发行业安全担忧。

    Source
  2. 科技执法

    英国警方借助Palantir(数据分析)AI破获80万英镑诈骗案

    2024年,Bedfordshire警方打掉跨国诈骗团伙“Fuck the Police”,该团伙在英国各地ATM作案3000余起,盗取80万英镑。警方缴获24部手机,数据量1.4TB,约等于50万本电子书。Palantir的AI系统“Nectar”在一天内完成逾10万条罗马尼亚语信息翻译,绘制嫌疑人关系图并分析行踪,若人工翻译需花费约3万英镑并耗时数周。2024年11月,六名男子被判刑。Palantir与NHS及英国国防部的合同金额超过5亿英镑,其AI工具已供11支警察部队使用;政府上月承诺投入逾1.15亿英镑,向全国43支警队推广AI,并设立Police.AI国家中心。

    Source
  3. 开源工具

    Composio 开源智能体编排器

    Composio 开源 Agent Orchestrator,帮助 AI 开发者构建可扩展的多智能体工作流,突破传统 ReAct 循环的局限。

    Source

+7 more headlines

  • 谷歌发布Gemini 3.1 Pro,业界关注DeepSeek回应
  • 利用视觉-语言模型扩展数据标注,为实体AI系统提供动力
  • 工程师一小时交付生产级SaaS
  • GLM-5与Qwen3.5发布,对标美国顶级模型
  • Uber想成为自动驾驶出租车的瑞士军刀。
  • 《商业时报》:塑造以人为本的AI未来
  • Meta(元宇宙)AI安全研究员称OpenClaw代理删光其收件箱
Unlock all 10 headlines + deep analysis →Free 3-day trial · cancel anytime
Browse all past briefings →