A hacker exploited Anthropic PBC’s artificial intelligence chatbot to carry out a series of attacks against Mexican government agencies, resulting in the theft of a huge trove of sensitive tax and voter information, according to cybersecurity researchers.The unknown Claude user wrote Spanish-language prompts for the chatbot to act as an elite hacker, finding vulnerabilities in government networks, writing computer scripts to exploit them and determining ways to automate data theft, Israeli cybersecurity sta ...

谷歌再次刷新文生图榜单。 去年8月，谷歌发布了Gemini图像模型Nano Banana，一度全网刷屏，成为现象级产品，同年11月，谷歌又发布了Nano Banana Pro，提供更高级的智能功能和 工作室级别的创意控制。 北京时间2月27日凌晨，谷歌又更新了，这次是Nano Banana 2（Gemini 3.1 Flash Image），兼具了速度和Pro版的性能，同时价格也更便宜了。谷歌表示，这 是团队目前最好的图像生成和编辑模型。 | Current models | All models | All | Open weights | First party foundation models | All | Global Leaderboard | Personal Leader | | --- | --- | --- | --- | --- | --- | --- | --- | | 11 | Creator TJ | | Model 17 | | ELO TT | 95% Cl | Samples TJ | | 1 | G Google | | | Nano Banana 2 (Gemini 3 ...

证券研究报告·行业深度报告·电子 [Table_Tag] [Table_Summary] 投资要点 ◼ 端侧模型牵引硬件重构：算力、存力与散热协同升级。从整机 AI 功能 看，2024 年行业整体仍以高频刚需场景为切入点，重点围绕图像消除、 文本摘要等低门槛功能；进入 2025 年，厂商明显加速向多模态创作能 力延展，覆盖语音、生成式图像等更复杂交互形态，并进一步向操作系 统底层渗透。整机 AI 竞争正从功能数量比拼，转向多模态体验与系统 级整合深度的综合较量。在整机级 AI 能力向多模态等方向升级的背景 下，端侧核心部件也正围绕内存与功耗等制约端侧体验的关键变量上进 行新一轮升级。在存储侧，三星 LPDDR6 产品在支持更高数据传输速率 和内存带宽的情况下，还从电路架构到电源管理进行了系统性重构，使 LPDDR6 在保持高速性能的同时，实现较上一代约 21%的能效提升。在 散热侧，三星于 2025 年 12 月 19 日发布 Exynos 2600 芯片，首次在移 动 SoC 中引入 High-k EMC 材料优化热传输路径，使热阻较 Exynos 2500 降低约 16%。在重载场景（如游戏与端侧 AI ...

OpenAI把自己的算力建设支出目标，从1.4万亿美元，直接砍到了6000亿美元。 你没看错，一刀下去，砍掉了8000亿美元。 这是什么概念？8000亿美元，比全球绝大多数国家一年的GDP还多。说扔就扔了，说不要就不要了。 这背后到底发生了什么？ 1.泡沫，终于开始漏气了 说真的，我没想到这一天来得这么快。 就在所有人还沉浸在"AI改变世界"的狂欢里，全球最炙手可热的AI公司，突然踩了一脚急刹车。 所以这次下调，说白了就是——吹出去的牛，终于兜不住了。 2.投资者不是傻子 有一个细节特别值得玩味。 据报道，OpenAI这次调整支出目标，是在"告知投资者"的过程中提出的。换句话说，不是OpenAI自己 突然良心发现，而是投资者施压了。 还记得几个月前吗？奥尔特曼站在聚光灯下，意气风发地宣布：我们要投1.4万亿美元搞基础设施建 设！ 1.4万亿，这个数字一出来，整个科技圈都沸腾了。媒体争相报道，投资者热血沸腾，仿佛AI的黄金时 代已经唾手可得。 但现实是什么呢？ 现实是，OpenAI去年一整年赚了131亿美元，听起来不少对吧？但同时烧掉了80亿。一个毛利率只有 33%的公司，张口就要砸1.4万亿——说难听点，这叫 ...

在这个被称为"AI兑现期"的关键分水岭，资本市场变得极其现实。过去几年，只要科技巨头们抛出千亿 级的资本开支计划，讲述一个改变世界的通用人工智能愿景，华尔街就会毫不吝啬地给出高估值。但如 今，风向变了。市场不再为单纯的"投入规模"买单，而是拿着放大镜寻找真金白银的回报率。 当时间指针拨入2026年，全球科技圈的空气中弥漫着一种微妙的焦灼感。 强如OpenAI，尽管其ChatGPT每周服务超过8亿用户，2025年斩获了130亿美元的惊人收入，但高达80亿 美元的现金亏损依然让其深陷"高收入、高亏损"的泥潭。海外巨头们正集体面临一个尴尬的"投入陷 阱"：巨额资本开支疯狂吞噬当期利润，但AI能力却难以深入用户高频场景，变现路径依然停留在卖算 力或勉强嵌入广告的初级阶段。 就在全球市场对AI回报周期产生集体质疑之时，作为中国乃至全球AI引领者的百度，交出了一份堪 称"破局"的答卷。根据其发布的2025年第四季度及全年财报，百度2025年总营收达到1291亿元，其中AI 业务营收高达400亿元。而在第四季度，百度总营收为327亿元，AI业务收入已经占到百度一般性业务 收入的43%，远超市场预期。 在这场全球AI价值变现 ...

Core Points - The CoreWeave class action lawsuit has been initiated against CoreWeave, Inc. and certain executives for alleged violations of the Securities Exchange Act of 1934 during the class period from March 28, 2025, to December 15, 2025 [1] - CoreWeave is positioned as an AI cloud computing company and had announced a significant deal worth up to $11.9 billion with OpenAI shortly before its IPO [2] - The lawsuit claims that CoreWeave overstated its ability to meet customer demand and failed to disclose risks associated with reliance on a single third-party data center supplier, which could negatively impact revenue [3] Financial Impact - On October 30, 2025, Core Scientific announced it did not receive enough shareholder votes to approve its merger with CoreWeave, leading to a more than 6% drop in CoreWeave's share price [4] - Following a revenue guidance reduction on November 10, 2025, due to delays from a third-party data center developer, CoreWeave's shares fell over 16% [5] - A December 15, 2025 article revealed further delays in data center delivery, causing an additional 3.4% decline in CoreWeave's share price [6] Legal Process - Investors who purchased CoreWeave securities during the class period can seek appointment as lead plaintiff in the class action lawsuit, representing the interests of all class members [7] Firm Background - Robbins Geller Rudman & Dowd LLP is a leading firm in complex class action litigation, having recovered over $916 million for investors in 2025 and a total of $8.4 billion over the past five years [8]

Core Insights - Google has launched Nano Banana 2, which emphasizes "speedy experience" and "professional image quality," with a significant new feature of "real-time connectivity" that enhances its capabilities beyond mere image generation [1][10]. Group 1: Product Features - Nano Banana 2 integrates with Gemini's search capabilities, allowing the model to understand, retrieve, and generate images that are more aligned with real-world information structures [1]. - The model can generate detailed street scenes and character interactions that are nearly indistinguishable from real photographs, showcasing its advanced rendering capabilities [2][3]. - The "real-time connectivity" feature allows for precise generation of images based on real geographical and meteorological data, enhancing the model's utility in various contexts [5][41]. Group 2: Competitive Landscape - In the latest Artificial Analysis rankings, Nano Banana 2 secured the top position, with its image editing capabilities ranking third, while being priced at half of its closest competitor, OpenAI [8][9]. - The competition in the image generation sector has intensified, with leading models showing minimal score differences, indicating a close race among top players [9]. Group 3: User Experience and Applications - Users have reported that Nano Banana 2's ability to generate high-quality images with accurate text rendering has significant implications for marketing materials and global communication [45]. - The model's enhanced consistency in character design and scene elements allows for seamless storytelling in comics and branding [51]. - The ability to visualize complex concepts and data efficiently positions Nano Banana 2 as a transformative tool in education, research, and data analysis [43][42]. Group 4: Technical Upgrades - The model has improved text rendering and translation capabilities, allowing for natural integration of text within images, which is crucial for marketing and promotional content [45]. - It supports multiple resolutions, including a new 512px option optimized for low-latency scenarios, making it suitable for rapid prototyping and iteration [64]. - The visual quality of generated images has been upgraded, with more natural lighting, richer materials, and sharper details, making it a viable tool for professional use [66].

此前，阿里首款自研AI眼镜夸克2025年11月27日正式发布，深度搭载千问AI助手。功能上，该产品深 度融合支付宝、高德地图等阿里生态核心场景，支持导航、支付、实时翻译等实用功能，吸引不少消费 者排队到线下门店体验。 图片来 源：视觉中国 2月27日，千问将在西班牙巴塞罗那举行的2026年世界移动通信大会（MWC）上发布首款同名AI眼 镜，并于3月2日开启线上线下全渠道预约。 智通财经了解到，千问APP点外卖、打车等能力，也将集成到千问AI眼镜等终端设备。据阿里内部人士 透露，除AI眼镜之外，千问还会在年内陆续发布AI指环、AI耳机等产品，并面向全球市场发售。 2月23日，千问曾发布数据，春节期间，用户在千问上"一句话下单"近2亿次，有超过400万的60岁以上 用户让千问帮忙下单。不限于手机屏幕，阿里也希望未来将这些能力无缝延伸至AI眼镜等硬件产品 中，实现从"指尖交互"到"场景化交互"的跨越。 去年12月，阿里成立千问C端事业群，由阿里巴巴集团副总裁吴嘉负责。据悉，该事业群由原智能信息 与智能互联两个事业群合并重组而来，包含千问APP、夸克、AI硬件、UC、书旗等业务。 千问的硬件布局，并非个例。当前，全球 ...

格隆汇2月27日｜五角大楼要求Anthropic按"所有合法用途"授权Claude，Anthropic因担忧"大规模监控"和"完全自主攻击"拒绝妥协，称"无法昧着良心答应"。 军方威胁若周五前不答应将撕毁2亿美元合同，而马斯克旗下的xAI已同意该条款。值得注意的是，在模拟战争游戏中，顶级大模型95%的情况下会建议使用 核武器。 200多名谷歌和OpenAI的员工签署公开信，要求自家公司与Anthropic一道，"搁置分歧，团结一致，继续拒绝美国防部提出的使用模型进行国内大规模监控 和在无人类介入的情况下执行自主杀人的要求"。 SIGNATORIES Google: 168 signatories OpenAl: 46 signatories Current employee, Google, Staff Research Scientist Rif A. Saurous Current employee, Google, Principal Research Scientist Kevin Slagle Current employee, OpenAl, Research Scientist Tom Scha ...

南方财经 21世纪经济报道记者吴斌 报道 随着华尔街对AI热潮是否可能演变为泡沫的担忧加剧，英伟达的惊艳业绩未能打消投资者的担忧。 在美国东部时间2月25日的盘后交易中，英伟达股价最初上涨4%，随后回吐涨幅，并最终在26日收跌 5.45%。 在大跌之前，英伟达给出的第一财季业绩展望轻松超出分析师平均预期，且第四财季营收大幅增长 73%。 英伟达炸裂财报未能重振AI交易，26日纳指大跌逾1%。标普500指数收跌0.54%，报6908.86点；纳指跌 1.18%，报22878.38点；道琼斯工业平均指数涨0.03%，报49499.2点。 但分析师仍越来越担心英伟达依赖少数几家被称为"超大规模云服务商"的大型数据中心运营商，以及 OpenAI等AI初创企业。Visible Alpha研究主管Melissa Otto表示，英伟达面临"很多担忧"，涉及其所依赖 的大型科技公司和初创企业客户如何为数千亿美元的AI基础设施支出提供资金。 微软、谷歌、亚马逊和Meta今年合计预计资本支出为6600亿美元，主要集中在AI数据中心。这些科技 巨头现在面临着在烧钱如山的情况下进入债券和股票市场融资的前景。 英伟达缘何陷入财报下跌" ...