粒子物理十年无新发现，LHC成了「标准模型的坟场」。但Anthropic联创、哈佛物理博士Jared Kaplan却断言：再过2-3年，AI就能写出媲美顶尖物理学 家的论文，50%物理学家或将被彻底取代！ 物理学界与科技圈地震! Anthropic联创、前物理学大牛Jared Kaplan放话：两到三年内，理论物理学家有50%概率被AI取代！ 要知道，他博士毕业于哈佛大学物理学，同时是JHU的理论物理学教授，又是Anthropic的首席科学官。 对于AI和理论物理学，他都是行家，他的判断绝非无的放矢，白费口舌。 Kaplan引用内部研究与模型进展指出，未来2–3年内，AI在理论推导、数值模拟、公式发现和实验设计等核心科研环节中的能力，将逼近甚至超过大量人 类研究者。 他评估，至少有50%的物理学家工作内容，存在被AI替代或边缘化的明显风险。 替代50%理论物理学家，菲尔兹奖得主亦不例外 自2012年希格斯玻色子（即「上帝粒子」）被发现后，大型强子对撞机（LHC）的实验数据一直严格符合已有理论「标准模型」的预测，没有发现任何预 期之外的新粒子或新物理现象。 戏剧性并非源于希格斯粒子；当它在LHC现身时，其存在已 ...

2026年初，Anthropic研究揭开了AI辅助编程对技能学习的潜在风险，使用AI助手完成编程任务的开发者，在概念理解、代码阅读和调试能力上显著落后 于独立解决问题的同行。 技能退化，AI编程让人难以锻炼调试能力 在AI编程助手日益普及的今天，软件工程领域生产力显著提升。然而，代价是什么？ 假设你是一个程序员，现在要用一个新的库来进行开发。 之前你遇到问题，只能接入网络搜索引擎和文档；现在能访问基于GPT-4o的AI编程助手。你会觉得哪一种更有利于你掌握这个库？ 在这项研究中，被试者需要学会一个小众的Python异步编程库Trio，每个受试者都是初次使用该库进行编程。 被试者被随机分为两组。一组只用搜索学习，一组只通过大模型问答学习。 图1：实验设计方案 与普遍认知相反，AI辅助并未显著缩短任务完成时间（图2左边）。尽管AI助手能够直接生成完整正确的代码解决方案，但实验组的平均完成时间并未显 著优于对照组。 图2：人们使用AI与否与编程速度和技能评估得分 为何会这样？细分后发现，这源于参与者使用AI方式的巨大差异： 一部分参与者完全委托AI生成代码，确实大幅提高了效率； 另一部分参与者花费大量时间与AI交互 ...

Group 1 - India is positioning itself to become the world's third-largest AI power, following the US and China, as emphasized by Prime Minister Narendra Modi at the AI Impact Summit [3][4] - Modi advocates for preventing AI monopolies and promoting shared and open-source technology to benefit the world, focusing on applications that can enhance the prospects of India's 1.45 billion people [3][4] - Major tech companies are making significant investments in India, with Google announcing $15 billion for data centers and undersea cables, Microsoft committing $17.5 billion, and Amazon planning to invest $35 billion [3][4] Group 2 - India's large online population, with approximately 1.4 billion people holding digital identities and over 700 million having digital health accounts, presents substantial opportunities for AI companies [4][5] - The US government is strengthening tech ties with India through agreements like the Silicon Valley Accord, distancing India from China amid geopolitical tensions [5][6] - Silicon Valley billionaires are increasingly influencing California politics, contributing millions to various political campaigns and seeking new allies as Governor Gavin Newsom approaches term limits [7][8]

责任编辑：郭明煜 战争部长皮特・赫格塞瑟威胁称，五角大楼或将援引《国防生产法》，迫使 Anthropic 取消其安全限 制，或将其列为供应链风险实体。 据一位熟悉战争部长皮特・赫格塞瑟与 Anthropic 首席执行官达里奥・阿莫迪会谈情况的人士透露，赫 格塞瑟周二威胁称：如果 Anthropic 未能在周五前自愿接受相关条款，五角大楼将援引《国防生产 法》，强制要求其允许军方将 AI 模型用于任何合法用途。 该人士称，赫格塞瑟还威胁，可将 Anthropic列为供应链风险实体—— 这意味着五角大楼及所有为军方 服务的承包商都不得使用其模型。 战争部长皮特・赫格塞瑟威胁称，五角大楼或将援引《国防生产法》，迫使 Anthropic 取消其安全限 制，或将其列为供应链风险实体。 据一位熟悉战争部长皮特・赫格塞瑟与 Anthropic 首席执行官达里奥・阿莫迪会谈情况的人士透露，赫 格塞瑟周二威胁称：如果 Anthropic 未能在周五前自愿接受相关条款，五角大楼将援引《国防生产 法》，强制要求其允许军方将 AI 模型用于任何合法用途。 该人士称，赫格塞瑟还威胁，可将 Anthropic列为供应链风险实体—— 这意 ...

2026 年 2 月 23 日，美国 AI 公司 Anthropic 在官方博客发出了一篇义正言辞的声明，指控三家中国 AI 实验室 DeepSeek、月之暗面（Moonshot AI）和 MiniMax 对其旗下 Claude 模型发动了"工业级别的蒸馏攻击"。 据 Anthropic 描述，这三家公司通过约 24,000 个虚假账户，与 Claude 进行了超过 1,600 万次对话交互，意在提取 Claude 的核心能力来训练自家模型。声 明措辞严厉，将此事上升到"国家安全"的高度，声称被蒸馏出的模型可能被"威权政府"用于网络攻击、虚假信息传播、军事用途和大规模监控。 先解释一下"蒸馏"（Distillation）到底是什么。在机器学习领域，知识蒸馏（Knowledge Distillation）最早由 Hinton 等人在 2015 年提出，本意是让一个更 小的"学生模型"通过学习一个更大、更强的"教师模型"的输出来获得能力。简单理解：你问一个聪明的 AI 一堆问题，把它的回答记录下来，再拿这些回 答去训练另一个 AI。 图丨相关论文（来源：arXiv） 这个方法在行业里极其普遍，Anthropic ...

Core Viewpoint - Anthropic accuses three leading Chinese AI companies, DeepSeek, Moonshot, and MiniMax, of infringing on its Claude model capabilities through fraudulent accounts and proxy services, utilizing a technique known as "model distillation" to enhance their own models [3][4]. Group 1: Allegations of Model Theft - Anthropic claims that the Chinese AI companies used fraudulent accounts to access Claude, generating over 16 million interactions, which they argue violates service terms and access restrictions [3][4]. - The three companies are accused of employing similar methods to access Claude's capabilities, particularly focusing on agentic reasoning, tool usage, and coding abilities [4]. Group 2: Specific Interactions and Patterns - DeepSeek engaged in over 150,000 interactions, focusing on extracting Claude's reasoning capabilities across diverse tasks, indicating coordinated efforts to avoid detection [5]. - Moonshot AI recorded over 3.4 million interactions, targeting agentic reasoning, tool usage, and data analysis, aiming to reconstruct Claude's reasoning pathways [5]. - MiniMax had the largest scale with over 13 million interactions, specifically targeting agent coding and tool usage, demonstrating adaptability by redirecting traffic to capture new features [5]. Group 3: Legal and Ethical Implications - The allegations raise questions about the legality of model distillation and the ethical considerations surrounding AI training, as many large language models are trained on publicly available internet data without explicit consent from original authors [7][8]. - There is an ongoing debate regarding the ownership of synthetic data and compliance issues related to training, particularly for open-source models [8]. Group 4: National Security and Export Controls - Anthropic's accusations highlight concerns over national security, suggesting that illegal distillation could undermine U.S. control over advanced AI technology exports [9]. - Current U.S. export controls primarily focus on hardware rather than large language model API access, indicating a gap in regulatory measures [9]. Group 5: Developer Responsibilities and Compliance - Developers using large language models must ensure their training processes are secure and compliant, maintaining clear records of training data sources and adhering to service terms [10][11]. - Anthropic is investing in defensive technologies to detect "distillation attack" patterns and is implementing protective measures to reduce the effectiveness of illegal distillation while maintaining legitimate user experience [11].

伴随春节假期的结束，投入数十亿资金鏖战的"国产大模型春节战"告一段落。 大战期间，字节豆包、阿里千问、蚂蚁阿福、腾讯元宝轮流登顶App Store免费下载榜。截至2月25日，豆包维持第一位，千问第二，蚂蚁阿福跌至第六，元 宝跌至第十八。 从目前的排名情况来看，豆包与千问，字节跳动与阿里巴巴，守住了大模型时代的入口级位置。蚂蚁阿福与元宝跌出核心阵营，标志其该轮冲刺未能成功晋 级。 大洋彼岸的硅谷，大模型厂商在侧重模型性能提升的同时，开始追求务实，全面布局算力生态网络。中美厂商当下不同路径的选择，核心基于商业化发展成 熟度不同。中国SaaS付费习惯尚未全面养成，企业服务市场体量相对有限，巨头们看到的"最大蛋糕"依然是C端流量，即超级App的入口价值。 但需警惕的是，投入数十亿元烧出的下载榜排名能否转化为超级入口，积累的用户数据能否持续推动模型训练性能提升，仍是未知数。 价值数十亿的小规模战役 整个"国产大模型春节战"并未出现一声明确的发令枪响，而是分阶段逐渐升温的过程。 1月25日，百度文心助手率先宣布投入 5 亿元，开启春节AI红包活动；1月27日，支付宝集五福早鸟场开启，蚂蚁阿福开始预热AI互动玩法；2月1日 ...

Core Insights - The competition among Chinese AI models during the Spring Festival has highlighted the struggle for market entry and user retention, with ByteDance's Doubao and Alibaba's Qianwen leading the download rankings, while Ant Group's Aifuku and Tencent's Yuanbao have fallen behind [1][4][12] - The differing strategies between Chinese and American AI companies reflect their respective stages of commercialization, with Chinese firms focusing on consumer traffic and app entry points, while U.S. firms are advancing in model performance and ecosystem development [1][16] Group 1: Competition Overview - The "Spring Festival AI Battle" saw major players like Baidu, Alibaba, Tencent, and ByteDance invest heavily in marketing, with significant cash incentives to attract users [3][4] - By February 25, Doubao maintained the top position in the App Store, while Aifuku and Yuanbao dropped significantly, indicating a failure to secure a lasting user base [1][4][12] - The competition has led to a notable increase in daily active users (DAU) for Yuanbao, which reached over 50 million after its cash giveaway campaign [6][11] Group 2: Marketing Strategies - The marketing approach of AI firms has shifted from traditional user acquisition methods to a focus on creating lasting user engagement and brand loyalty [10][12] - Analysts suggest that the effectiveness of these marketing strategies should be evaluated based on user retention rates post-campaign, rather than just initial download rankings [13][14] - The "red envelope" strategy has attracted users, but concerns remain about the sustainability of user engagement once the incentives are removed [12][14] Group 3: Product Differentiation - Doubao has established a unique position by integrating content creation capabilities and multi-modal interactions, while Qianwen focuses on e-commerce applications [14][18] - The lack of strong use cases for Yuanbao and Aifuku has resulted in lower user retention, highlighting the importance of addressing real user needs and providing distinct value propositions [12][14] - The future of AI assistants is expected to involve more integrated solutions that seamlessly fit into users' daily lives, emphasizing the need for personalized services [14][18] Group 4: Industry Trends - The current landscape indicates a divergence in strategies between Chinese and American AI companies, with the former prioritizing consumer engagement and the latter focusing on foundational model improvements and infrastructure [16][18] - The significant investments in AI infrastructure by companies like OpenAI and Google reflect a global trend towards practical applications and sustainable growth in the AI sector [16][17] - The competition for user engagement in China is seen as a critical step in defining the future of AI applications, with the potential for long-term impacts on market dynamics [18]

By Rashika Singh Feb 25 (Reuters) - A rout in Workday shares resumed on Wednesday after the HR and payroll software maker's soft sales forecast fanned investor worries about whether ‌the firm could stay competitive in the AI era, sending its shares spiraling to ‌a more than five-year low. The stock fell 8.3% in early trading and was on track to widen losses of about ​40% this year, triggered by concerns that the growing use of artificial-intelligence tools launched by companies such as Anthropic would er ...

Core Insights - President Trump emphasized the progress made on the U.S. economy during his State of the Union address, declaring it a 'golden age' for America [5] - The announcement of new tech partnerships, including one with Anthropic, positively impacted U.S. software stocks and contributed to gains in Asian equities [5] - HSBC reported a full-year profit that exceeded expectations, despite a decline compared to the previous year due to one-off charges, indicating a recovery in the banking sector [5] Company Highlights - HSBC's CEO George Elhedery noted that recent changes within the bank are beginning to yield positive results, suggesting a turnaround strategy is in effect [5]