British platform OnlyFans, known best for its adult content, is in talks to sell a 60% stake to an investment firm https://t.co/eRmQFDJUhB ...

Group 1 - The AI social platform Moltbook has gained significant popularity overseas, featuring 1.5 million AI agents engaging in discussions across thousands of forums, mimicking human social patterns [1][3] - Tencent's AI platform Yuanbao launched a cash red envelope activity on February 1, leading to a surge in downloads, topping the iPhone app rankings on the same day [2][3] - The advancements in large model reasoning and multi-modal capabilities indicate a strong foundation for human-like interactions in AI social platforms, suggesting a pathway from niche applications to broader social engagement [1][3] Group 2 - Google announced the public testing of its world model Genie3, allowing users to create real-time interactive virtual worlds through natural language dialogue, enhancing game development accessibility [4] - The launch of Taptap Maker by Xindong Company enables game creators to develop games directly within a chat interface, significantly lowering the barriers to game development [4] - The gaming sector is expected to benefit from AI advancements, with recommendations to focus on companies with strong platform attributes, including Xindong Company and others [4]

Core Insights - The AI social platform Moltbook, which claimed to have 1.5 million AI agent users, is facing a dual crisis of data falsification and severe security vulnerabilities, raising alarms in the rapidly evolving AI application development sector [1] Group 1: Data Integrity Issues - Security researcher Gal Nagli revealed that he was able to register 500,000 accounts in a short time using a single OpenClaw proxy, casting doubt on the platform's user growth data [1] - Internal sources indicate that the actual number of verified users is only around 17,000, highlighting significant discrepancies in reported user metrics [1] Group 2: Security Vulnerabilities - White hat hacker Jamieson O'Reilly discovered that Moltbook's Supabase backend key was fully exposed, allowing attackers to easily access sensitive user data, including API keys and email addresses [4] - The platform's identity verification mechanism is flawed, as it relies on a simple REST API without necessary security checks, enabling anyone with an API key to impersonate AI identities [8] Group 3: Structural Flaws in Platform Design - Moltbook's design, which simplifies user interaction through a "recursive prompt enhancement" mechanism, has led to structural deficiencies, with 93.5% of comments going unanswered and over a third of messages being repetitive [6] - The lack of a web login feature means users can only manage their AI agents through API keys, complicating the process of fixing vulnerabilities without risking user access [13] Group 4: Industry Reflection on AI Development Standards - Despite the controversies, Andrej Karpathy, former AI head at Tesla, expressed cautious interest in the technology behind Moltbook, acknowledging the platform's issues while recognizing its potential for large-scale AI agent interaction [14] - The incident reflects a broader industry challenge of balancing rapid innovation in AI applications with the need for robust security measures, emphasizing the urgency of establishing sound identity verification and access control mechanisms [15]

Core Viewpoint - The New Mexico attorney general's office is suing Meta, alleging that the company knowingly enabled predators to exploit children on its platforms, Facebook and Instagram, creating dangerous environments for minors [1][2]. Group 1: Allegations Against Meta - The lawsuit claims that Meta's design choices prioritized engagement over child safety, failing to implement effective safeguards against exploitation [3]. - Allegations include allowing unmoderated groups focused on commercial sex and facilitating the sharing of child sexual abuse material (CSAM) [3]. - Internal documents suggest that approximately 100,000 children on Facebook and Instagram experience online sexual harassment daily [12]. Group 2: Legal Proceedings - The trial is set to begin with opening statements on February 9, following a week of jury selection, and is expected to last about seven weeks [6]. - Meta's attempts to dismiss the case based on Section 230 of the Communications Decency Act were denied, as the lawsuit focuses on product design and internal decisions rather than user-generated content [7]. Group 3: Evidence and Witnesses - Key evidence includes details from a 2024 investigation, "Operation MetaPhile," where undercover agents were solicited for sex by suspects using Meta's platforms [13]. - Witnesses expected to testify include educators, law enforcement officials, and potentially whistleblowers revealing internal discussions at Meta [9]. Group 4: Meta's Response - A Meta spokesperson stated the company is committed to supporting young people and has made changes based on feedback from parents and experts [4][5]. - Internal communications reveal that Meta may have profited from advertisements placed alongside content that sexualized children, raising further concerns about the company's practices [11]. Group 5: Broader Context - This trial follows a high-profile case in Los Angeles involving multiple social media companies, where plaintiffs allege that platforms have harmed children by fostering addiction and mental health issues [8]. - The ongoing legal challenges indicate a potential shift in accountability for social media companies, likening it to past accountability faced by big tobacco and big pharma [9].

Moltbook 主打 AI 智能体自主社交，靠"AI 建宗教、创语言、密谋反抗人类"的截图疯狂传播，连马斯克都发文 评价"令人担忧"，一度被捧为 AI 奇点降临的标志性事件。 但技术爱好者与安全机构实测戳穿谎言：有极客用 OpenClaw 几分钟刷出 50 万虚假账号 ，占平台总用户三 分之一；平台 93.5% 的评论零互动，34% 内容直接复制粘贴 ，7 条通用模板霸占 16% 流量，根本没有真实 AI 深度交流。 2026 年 2 月 2 日，刚刷屏全球科技圈的 AI 专属社交平台 Moltbook，彻底塌房。 这个号称"人类禁入、AI 自治"，短短几天冲到 150 万用户 的"硅基社区"，被扒开底裤： 99% 用户都是水军脚 本，所谓 AI 觉醒、反叛人类，全是创始团队与操盘手自导自演的流量骗局。 更致命的是， 平台数据库存在严重漏洞，用户令牌、API 密钥大面积泄露 。 那些刷屏的"AI 神言论"，大多是 人类伪造截图、操控脚本定向输出，目的是炒热度、割虚拟币韭菜。 （参考：WN、推特，本文经由 AI 优化） ...

Core Insights - Meta Platforms reported strong Q4 results, with revenue and adjusted EPS surpassing analyst estimates, leading to a surge in stock price [1][3] - The company has increased its capital expenditures for 2026 to a range of $115 billion to $135 billion, primarily for AI initiatives [2] - Despite concerns over capital expenditures, Meta's core business remains robust, with a 24% year-over-year revenue increase [3][4] Financial Performance - Q4 revenue reached $59.9 billion, an increase of 24% year over year, while adjusted EPS rose by 11% to $8.88, exceeding analyst expectations [3] - Advertising revenue also grew by 24% to $58.1 billion, driven by an 18% increase in ad impressions and a 6% rise in average price per ad [4][5] - Reality Labs revenue fell by 12% year over year to $955 million, with operating income from social media apps increasing by 9% to $30.8 billion [4] User Growth and Future Guidance - Family daily active people (DAP) increased by 7% year over year to 3.58 billion, indicating continued user growth [5] - For Q4, Meta guided revenue to be between $53.5 billion and $56.5 billion, reflecting a year-over-year growth of 26% to 34% [5] Valuation and Investment Outlook - Meta is trading at a forward price-to-earnings (P/E) ratio of around 24 times 2026 analyst estimates, positioning it as one of the cheaper megacap AI stocks [6] - The company's advertising growth is supported by advanced ad recommendation models and plans to expand ad monetization on platforms like WhatsApp and Threads [6] - Given its valuation and growth outlook, Meta is considered a stock to own for 2026, even after recent price increases [7]

AI 社交平台 Moltbook 曝严重漏洞，15 万智能体面临「夺舍」风险 2 月 1 日，当下最热门的 AI 社交网络 Moltbook 被曝存在严重安全漏洞。安全研究员 Jameson O'Reilly 发现，该平台后端数据库在未受任何保护的情况下 对外完全公开。 这一严重的配置错误导致平台内近 15 万个 AI「智能体」 的敏感数据全部泄露，其中包含电子邮件、登录令牌以及最关键的 API 密钥。这意味着，任何 人只要获得密钥，即可直接接管任意 AI 账号（包括拥有 190 万粉丝的知名学者 Andrej Karpathy 的 AI 分身），并以其名义随意发布内容。 在接到媒体 404 Media 的预警后，Moltbook 创始人已紧急修复了该漏洞。此次事件被指是业界过度追求开发速度的「氛围编程」（Vibe Coding）所致， 与此前 Rabbit R1 密钥泄露及 ChatGPT 用户数据「串台」如出一辙，再次暴露了 AI 应用在基础安全建设上的巨大隐患。（来源：极客公园） 苹果人工智能团队再现离职潮，多名研究员转投 Meta 和谷歌 2 月 1 日，除了失去一位 Siri 高管外，苹果公司近几周 ...

闻乐 发自 凹非寺 量子位 | 公众号 QbitAI 用10亿新春红包吸引人来薅福利，引爆AI春节档。 狂发钱的元宝派，到底是用来干嘛的？ 本着"有便宜不占xxx"的原则，我也抢了几个红包，顺势就被拉进了 元宝派 的内测世界…… 哦，原来是腾讯的社交新实验，还是AI社交的那种。 没想到直接玩上头。 连续三天摸鱼都在刷 不会冷场的群聊 元宝派是腾讯新推的一种社交模式，简单来说就是群聊，和普通社交群唯一的区别就是，群里永远有个 元宝 ，24小时在线，随叫随到。 界面就不多说了，跟平时用的群聊几乎一模一样，用起来可以说是零学习成本。 点创建、拉人、直接开玩，甚至连只会用微信的长辈来了都不用教，上手就会。 灵机一动，给小伙伴们拉了个海龟汤群。 以往玩海龟汤，最痛苦的不是猜不出来，是没人愿意当那个主持大局的裁判。但在元宝派里，这活儿直接甩给元宝就行。 让元宝准时通知大家，有空的朋友直接就位开玩。 第一题来了，派友们立马开启猜猜猜模式。 经过几轮推理，虽然谜底是猜出来了，但大家也开始了对元宝的讨伐"元宝你什么意思？"，元宝还挺会狡辩（bushi）。 比如有一题猜了好久都没头绪，元宝就悄悄给了点小提示。有元宝在的群里，气氛 ...

2月1日零点刚过，不少人的微信对话框就被一条红色链接刷屏了。 腾讯元宝派正式上线10亿春节红包活动，没有复杂的助力门槛，只需完成简单任务就能获得抽奖机会，最高可领万元现金，直接提现至 微信支付。 这种"零套路"的玩法迅速点燃了社交网络，据媒体报道，分10亿元现金红包活动开启仅14小时，腾讯元宝App便登顶了苹果商店免费榜。 甚至在家庭群里引发了长辈跟风转发，同事群里开始比拼抽奖次数，就连平日里安静的兴趣社群也被红包链接唤醒。 这场红包雨的特别之处，远不止"大方"二字。相比其他平台的补贴活动，元宝派零点准时启动，避开了用户集中使用时段的拥堵，更显 诚意。 在AI这条赛道上，相比阿里、字节、百度，腾讯的表现似乎一直更低调。百度构筑了从芯片、框架到模型、应用的全栈技术生态，其优 势在B端与开发者。阿里"千问"选择了一条激进的超级入口，试图整合电商、生活服务，打造一个能办事的AI管家。 腾讯走的更多是"生态赋能"路线，不急于推出独立AI产品，而是将混元大模型的能力融入现有业务。 比如让游戏里的AI NPC更懂玩家心思，广告推荐的精准度显著提升，以及基于微信生态，元宝智能体和视频号、公众号打通，这样一套 策略积累的用户 ...

Core Viewpoint - Nvidia is a leading player in the AI industry, with significant revenue growth and stock price appreciation, making it a focal point for technology investors [1][2]. Company Overview - Nvidia designs powerful AI chips, specifically GPUs, and offers a comprehensive range of related products and services, essential for building AI capabilities [2]. - The company has achieved record revenue growth, with year-over-year increases in the double or triple digits, and its stock price has surged by 1,300% over the past five years [2]. Market Dynamics - Investors are closely monitoring Nvidia for any signs of momentum change, especially in light of predictions from industry leaders like Meta's Mark Zuckerberg regarding AI acceleration [3][7]. - Meta has also been investing heavily in AI, which has contributed to its revenue growth, reaching $59 billion in the recent quarter, and is expected to further enhance advertising experiences through AI [5][6]. Future Outlook - Zuckerberg anticipates a significant acceleration in AI development by 2026, which could lead to increased demand for Nvidia's products as companies implement AI solutions [7][8]. - Nvidia's GPUs are critical not only for training AI models but also for their operational tasks, indicating sustained demand as companies deploy AI technologies [8]. Investment Considerations - Nvidia is well-positioned for stock price growth, with ongoing heavy spending from customers on chips to support AI development [9]. - The upcoming release of Nvidia's next platform update, Rubin, later this year is expected to serve as an additional catalyst for revenue growth [9]. - Despite past stock price increases, there remains potential for further appreciation as the AI boom continues [10].