← AM·PM Brief

Meta Builds AI App Squad, Claude Code Talks: JustSayAI Daily Brief

· Morning brief · 10 news · 3:04

Audio in Mandarin Chinese · English transcript below

Meta’s AI SWAT, Claude speaks, China open-weights surge, deepfake wars, defense AI valuations double—sector scorching!

A core trend in today's AI landscape is that tech giants are rapidly integrating AI deeply into applications while simultaneously optimizing user experience and content governance. Meta's formation of a new applied AI engineering organization underscores its strategic direction of fully deploying AI across its products, while the launch of a voice mode for Claude Code and adjustments to ChatGPT's model behavior collectively point towards increasingly humanized and efficient AI assistant interactions. Notably, X platform's revenue-sharing policy for unlabeled AI-generated armed conflict videos also highlights the urgency of AI content ethics and platform responsibility.

Today's Top 3 Headlines

  1. 科技

    Meta将组建新应用AI工程组织

    Meta(元宇宙)计划组建一个全新的应用AI工程团队,该团队将采用扁平化架构,以加速AI技术在集团内部的深化与应用。

    Source
  2. 科技

    Claude(AI助手)Code上线语音模式

    Anthropic(AI公司)为Claude Code(AI助手)新增语音模式,工程师Thariq Shihipar周二在X宣布逐步上线,目前约5%用户可用,未来数周将扩大覆盖。用户输入/voice开启后可直接说出指令,例如“重构认证中间件”。公司未透露使用上限或是否采用第三方语音技术。Claude Code年运转收入已超25亿美元,周活跃用户自1月起翻倍。

    Source
  3. 技术教程

    使用 Unsloth 与 QLoRA 构建稳定高效的大模型微调流程

    教程展示如何使用 Unsloth 和 QLoRA 高效微调大型语言模型,构建稳定且端到端的监督微调流程。

    Source

+6 more headlines

  • ChatGPT(聊天机器人)的全新 GPT-5.3 Instant 模型将不再劝你“冷静点”。
  • X平台:未标注AI武装冲突视频将被暂停90天收益分成
  • 专家如何在深度伪造时代辨别真伪
  • “Gemini 3 Flash 击败 ChatGPT-5.2”的未证实报道引发关注
  • 中国实验室发布Qwen 3.5、GLM 5、MiniMax 2.5等旗舰开源模型
  • Anduril新一轮融资目标估值600亿美元
Unlock all 9 headlines + deep analysis →Free 3-day trial · cancel anytime
Browse all past briefings →