Tag
Meta announced it will release the final OS update (v213) for the Quest 2 and Quest Pro by March 2027, with security patches and critical fixes continuing through the end of 2027, as it shifts focus to Quest 3, 3S, and its upcoming VR Glasses.
S&P Global revised its outlook on Meta to 'negative,' citing negative free cash flow and surging AI capex projected to exceed $225B by 2027, highlighting broader doubts about the sustainability of the AI spending boom across hyperscalers.
The post argues Muse for SMBs is an underrated consumer agent story, highlighting Meta's 3.64B daily active users, 500M+ SMBs, connector ecosystem plans, Shopify integration, and 25% of Muse users completing purchases averaging over $200.
Meta's new TextSeal2 paper shows that blending two secret keys into one virtual key plus noise makes hidden watermarks in AI-generated text easier to detect, even with speculative decoding, reaching 89% detection at 0.1% false-positive rate versus TextSeal's 71%.
Discussion that Instinct is hitting a wall with WhatsApp, with some users reporting account bans, highlighting WhatsApp's massive distribution outside the US and China as a moat for Meta.
A new Meta and NYU paper shows that h1, a simple agent loop giving models fast Lean feedback and search tools, matches or beats hand-designed theorem-proving workflows: it outperformed Goedel-Architect on PutnamBench (598 vs 597 problems) at lower cost, and with Opus 4.8 solved 83/100 hard algebra problems versus 63 for Claude Code.
Meta is reportedly developing a humanoid robot prototype in collaboration with a Chinese startup, expanding its push into robotics beyond its wearables and AR efforts.
Meta reportedly built a 48 kg humanoid robot prototype called Max in just three months in partnership with Chinese robotics startup Fourier, with a second version already in development and no plans to sell the units.
前Meta研究员Stephen Roller在法庭证词中承认在职期间通过BitTorrent下载并分发了盗版Books3数据集，该数据集被用于训练Llama模型，尽管Meta此前否认通过BitTorrent获取该数据集。Meta回应称该下载可能是为个人用途，而原告计划要求法院进一步查明Meta获取和使用该数据集的详情。
Despite Meta's claims that its Muse agentic AI tool is driving significant shopping activity, executives across roughly 50 companies report minimal real-world impact from AI agents so far, with many optimizing their websites in anticipation of future adoption.
Canada's major banks — RBC, TD, CIBC, and BMO — announced they will block personal AI agents such as Meta's Muse and OpenAI's ChatGPT from accessing clients' online bank accounts, citing safety and security concerns amid reports of unauthorized third-party AI agent activity in accounts at major U.S. banks.
Meta 分享其 Catching JiTTests 方案在生产环境大规模部署的成果：利用 LLM 自动生成会失败的即时测试，在代码合入前捕获传统测试和 LLM 代码审查难以发现的 bug，并配套发布了详细论文。
A blog post reconstructs the 'Charter' referenced by Muse Spark Open in its chain of thought, analyzing Meta's model spec entirely from model outputs rather than leaked documents, with noted caveats about accuracy.
Princeton 与 Meta 的新论文提出 Self-Direction 方法：让失败的 AI 智能体反思并将教训训练成约 0.02% 大小的 LoRA 适配器，由路由器在遇到相似任务时自动加载，从而显著减少重复错误。在 Qwen3.6-27B 上，GAIA 任务成功率从 38.8% 提升至 55.8%，且跨 185 个早期错误的修复持久性远超普通重试。
Meta has quietly launched Style Sketch, an AI fashion app that builds a virtual closet from users' Instagram photos and suggests personalized outfits with virtual try-ons. The app is currently available only in Indonesia, Singapore, and Thailand, as Meta continues its broader push to release more AI-driven apps.
Meta更新商业用户服务条款，新增可抓取关联网站内容用于广告和商业产品的权利，虽删除了'训练'一词，但法律专家和新闻出版商团体警告该措辞仍可能为免费AI训练和抓取新闻内容开绿灯。
一篇技术博客深入解析 Meta 于 10 月 15 日发布的开源模型 Muse Spark Open（816B 总参数、42B 激活参数的 MoE 架构），并介绍 Daniel Gross 发起的 10 万美元 Open Kernel Challenge，邀请开发者在 H200 上优化其推理内核与 SGLang 服务性能。
Meta has launched Waddle World, an AI-powered Tamagotchi-like virtual pet app, currently available in the Philippines on iOS and Android, featuring journaling and mini-games while explicitly avoiding companion-chatbot design.
作者撰文指出科技公司在社交媒体与青少年心理健康研究中复制了烟草行业的拖延策略，并结合 Meta 内部诉讼文件披露其打压批评者、影响研究议程的做法，佐证了其团队关于科技公司介入相关研究的预印本论文。
Meta suspended the No Kings Coalition's Instagram account during a nationwide US voting campaign, first blaming high-frequency posting and then attributing the trigger to an automated tool using Claude AI reading profiles at 4x the normal rate. The incident raises concerns about tech companies deflecting accountability for political censorship by blaming AI automation.