此次一线员工的联合发声，直接挑战了科技巨头现有的合规与商业决策。资料显示，谷歌曾于2025年2 月撤销了其内部关于禁止将AI用于武器和监控的规定。目前的焦点在于，面对拥有近20万名员工的谷 歌和近1万名员工的OpenAI，这批签署者的行动能否实质性地迫使管理层重新确立AI技术授权的道德红 线。 凤凰网科技讯 2月27日，据ASIOS报道，谷歌与OpenAI的超200名员工近日联合签署了一封公开信，正 式声援Anthropic的立场，呼吁对先进人工智能（AI）在军事和监控领域的应用实施严格限制。这封联 名信明确反对将AI技术用于国内监控或全自动战争等争议性领域，以抵制美国五角大楼的相关技术诉 求。 截至当地时间周四下午，该信件已获得超160名谷歌员工和超40名OpenAI员工的核实签名，部分人员选 择匿名参与。信件指出，美国五角大楼正试图利用企业间的竞争，分别与谷歌和OpenAI进行谈判，以 获取此前被Anthropic明确拒绝的AI技术支持。对此，员工在信中呼吁两家公司的管理层搁置分歧，共 同抵制军方的此类要求，并强调整体行业应在技术底线上保持信息透明。 ...

Group 1: Paramount and Warner Bros. Acquisition - Paramount Skydance (PSKY) wins the bidding war for Warner Bros. Discovery (WBD) as Netflix (NFLX) decides not to raise its bid, deeming the revised offer from Paramount as "superior" [3] - The final purchase price from Paramount is $31.00 per share in cash, which values Warner Bros. at approximately $108 billion, reflecting a one-dollar increase from its previous bid [3] - Warner Bros. Discovery is contractually obligated to pay Netflix a breakup fee of $2.8 billion due to the termination of the original deal [4] Group 2: Anthropic and Pentagon Negotiations - Anthropic (ANTHRO) refuses the Pentagon's demand for unrestricted access to its AI models, stating that it cannot allow its technology to be used in all lawful cases without limitations [5] - The Pentagon has threatened to label Anthropic as a "supply chain risk" and may invoke the Defense Production Act to enforce compliance with its demands [7] - Anthropic signed a $200 million contract with the Department of Defense in July, becoming the first lab to deploy its models in classified mission workflows [8] Group 3: Burger King Whopper Update - Burger King is refreshing its iconic Whopper for the first time in nearly a decade, introducing a more premium bun and better-tasting mayo while maintaining the same beef and toppings [9][10] - The new Whopper will be served in a box instead of a wrapper to ensure quality upon delivery, reflecting changes made based on customer feedback [10] - Restaurant Brands International (QSR) announced the upgraded Whopper at its Investor Day event, with QSR shares closing up 3.3% [10]

2月26日，马年正月初十，广州开发区、黄埔区"新春第一会"会场内，广州开发区党工委黄埔区委副书 记、区长冼银崧戴着仅35克的智能眼镜走上台，用眼神翻页完成了脱稿演讲。 这款被誉为"下一代移动智能终端"的AI眼镜，由本土企业研发制造，背后串联着50多个核心零部件、超 30家制造商与服务商，勾勒出一条完整的产业协作链条。 但比这副眼镜更值得追问的是：当人工智能浪潮席卷而来，一个工业大区究竟需要什么，才能让这场变 革真正惠及企业？冼银崧给出的答案是——"资金无忧、算力无虑、场景无界、市场无限"。 这四个"无"，就像打开新大门的"四把钥匙。" 广州开发区党工委黄埔区委副书记、区长冼银崧特意佩戴了"黄埔造"的全球首款可全天候佩戴的智能眼 镜上台发言，向现场嘉宾推荐黄埔产品。通讯员 李剑锋 摄 算力的"平民化实验" "人工智能烧钱，烧得最多的就是算力。"会场外，一位企业负责人私下感慨。这是不争的事实。 AI大模型的训练成本动辄千万，对绝大多数中小企业而言，这是一道很难跨越的门槛。算力被称为AI 时代的"水电煤"，但问题是，水电煤的价格应该是普惠的，而算力还很贵。 黄埔的解法很直接：既然算力是新基建，那就把它当作基建来建— ...

Core Insights - The AI competition during the recent Spring Festival has significantly transformed social interactions, with major companies investing heavily to capture market share and educate users about AI [1][2][3] - The total investment from major players like Tencent, Alibaba, ByteDance, and Baidu exceeded 5 billion yuan, surpassing previous years' spending on similar promotional activities [1][2] Group 1: User Engagement and Market Dynamics - The engagement on social media platforms surged, with over 371 trending topics related to AI models on Weibo, accumulating 12.7 billion views and nearly 26.9 million interactions during the festival [1] - The rapid adoption of AI applications was evident, with significant increases in app downloads and user activity, such as the "Yuanbao" app reaching a peak of 1.2 billion in WeChat index within hours of its launch [3][5] - The competition among major players led to a dramatic increase in daily active users (DAU), with "Doubao" achieving 145 million, "Qianwen" at 73.5 million, and "Yuanbao" at 40.5 million during the festival [11] Group 2: Competitive Strategies and Outcomes - Companies aimed to replicate the success of past mobile payment wars by heavily subsidizing user acquisition, but the challenge remains whether these users will continue to engage post-subsidy [2][21] - "Qianwen" emerged as a dark horse, experiencing a 940% increase in user participation after launching its "Spring Festival Treat Plan," which significantly boosted its DAU [15][16] - "Doubao" maintained its lead due to effective marketing strategies, leveraging its parent company ByteDance's ecosystem for user acquisition and retention [15][16] Group 3: Industry Trends and Future Outlook - The AI industry is witnessing a clear bifurcation, with large companies focusing on user engagement and ecosystem integration, while smaller firms concentrate on technological advancements and B2B services [22][23] - The competition is expected to shift from casual interactions to solving real user needs, as companies like "Qianwen" and "Antifor" integrate AI into practical applications like e-commerce and health services [23][25] - The high operational costs associated with AI applications pose a challenge for sustainability, as the industry grapples with the balance between user growth and profitability [25]

Claude Code、ChatGPT Agent、Manus，再加上一堆大厂的企业级工作流平台，动不动就喊"替代打工人""创造万亿美 元价值"。麦肯锡的报告也在推波助澜：62% 的企业正在试水 Agent。 但把热闹先放一边，真正的问题是——在真实工作场景里，Agent 到底走到哪一步了？ 最近，MIT、哈佛、斯坦福等机构的研究团队发布了一份《2025 AI Agent Index》。 这帮学者干了一件很有价值的事： 进入2026 年，Agent 彻底火了。 他们把市面上最具代表性的30 个代理系统逐一拆开来看了一遍，并设计了45 个维度，仔细扒了扒这些产品的技术细 节、部署情况、设计架构、工具使用和安全机制等信息，得出了当下Agent最硬核的3个真相。 透过这部分报告，我们就能更清晰地了解当下Agent发展的真实情况。 01 产品形态收拢在这3个方向 大多数代理产品集中在2024–2025 年发布。 （蓝色柱状图代表与代理型人工智能产品相关的谷歌新增搜索词，红线则代表谷歌学术中包含"人工智能代理"或"代理 型人工智能"关键词的论文发表数量） 从产品形态看，AI代理基本收拢在了三个方向： 聊天式代理（12个 ...

Core Insights - DeepMind's latest paper introduces AlphaEvolve, which treats algorithm source code as a genome and uses Gemini as a genetic operator to evolve new game theory algorithms through a process akin to natural selection [1][5][20] - The evolved algorithms outperform human-designed optimal solutions in various tests, utilizing mechanisms that human researchers had not previously considered [1][22] Group 1: AlphaEvolve and Its Mechanism - AlphaEvolve is described as an evolutionary coding agent that uses the source code as a genome, with LLM acting as a genetic operator to mutate the code [5][20] - The process involves evaluating the fitness of each "offspring algorithm" based on its exploitability in a set of benchmark games, allowing the best-performing algorithms to survive and evolve further [5][6][20] Group 2: Target Algorithms and Historical Context - The focus of AlphaEvolve is on two core algorithm families in multi-agent reinforcement learning: Counterfactual Regret Minimization (CFR) and Policy Space Response Oracles (PSRO) [6][7] - Historically, researchers have manually tuned and designed variations of these algorithms, but AlphaEvolve automates this process, significantly expanding the search space for potential solutions [10][16][20] Group 3: Implications of AI-Driven Algorithm Design - The paper highlights that the design of game theory algorithms has traditionally been a challenging task due to the complexity of incomplete information games [12][13] - AlphaEvolve's approach allows for meaningful mutations in code, leading to the discovery of effective strategies that human experts had not conceived [17][25] - The results indicate a paradigm shift where AI not only executes algorithms but also invents them, achieving superior performance compared to human-designed methods [22][25] Group 4: Future Directions - DeepMind plans to apply this evolutionary framework to the complete design of deep reinforcement learning agents and explore mechanism discovery in cooperative games [25]

日前，国内大模型企业阶跃星辰可谓是动作频频，先是在今年一月宣布拿到50亿元B+轮融资的同时， 官宣旷视科技联合创始人、千里科技董事长印奇正式出任公司董事长，如今不过一个多月，市场传出阶 跃星辰正考虑在港交所IPO的消息。 据《财经》报道，阶跃星辰正在进行新一轮Pre-IPO融资，报道还称，阶跃星辰计划在今年6月30日前在 港股交表，预计今年年底完成上市。 截至发稿，阶跃星辰暂未回复相关传闻。 阶跃星辰上述消息引发广泛关注与其"大模型六小龙"之一的身份不无关系。就在今年1月初，"大模型六 小龙"中的智谱、MiniMax相继登陆港交所，两只股票在春节期间股价双双创新高，获得资本追捧，市 值也一度突破3000亿港元。尽管后期股价均有所回调，智谱、MiniMax毫无疑问仍然是二级市场中的明 星企业。 在AI热潮之下，阶跃星辰如何在竞争中站稳脚跟，又能否承接住智谱、MiniMax 打开的港股大模型估 值窗口？成为市场关注的焦点。 港交所AI大模型第三股要来了？ 以最近的一轮融资来看，阶跃星辰完成超50亿元B+轮融资，创下过去12个月中国大模型赛道单笔最高 融资纪录，投资方包括上海国投先导基金、国寿股权、浦东创投、徐汇资 ...

Core Insights - The departure of key members from the Grok team, including Jiayi Pan and Toby Pohlen, raises questions about the internal dynamics at xAI [1][3] - Jiayi Pan's journey from a novice to a core contributor at Grok 4 highlights a significant evolution in his expertise and approach to AI technology [4][7] Group 1: Jiayi Pan's Contributions - Jiayi Pan began his AI journey in 2019, studying computer science and electrical engineering at the University of Michigan, and graduated in 2023 [4] - He developed SWE-Gym, an environment that integrates reinforcement learning (RL) into software engineering, during his early projects at UC Berkeley [6] - Pan's work at xAI included optimizing the RL module for Grok 4, which advanced the model's capabilities from simple predictions to self-verification [7] Group 2: TinyZero Project - In 2025, Jiayi Pan announced the open-source TinyZero, a model with a training cost of only $30, achieving self-verification and reasoning capabilities through pure reinforcement learning [8][10] - TinyZero demonstrated significant improvements in task accuracy, with a model's performance on the Countdown task increasing from 0% to over 80% after RL training [9] - The project challenges the notion that advanced reasoning capabilities require massive infrastructure investments, as evidenced by the stalled Stargate project by Sam Altman [10] Group 3: Implications of TinyZero - TinyZero's self-correcting abilities, including generating intermediate thought processes during tasks, suggest a new frontier in AI development that does not rely on large-scale resources [12][15] - The combination of Jiayi Pan's projects indicates a potential for AI to not only correct itself but also to optimize its training processes, hinting at a form of "self-evolution" [16] - The emergence of affordable AI models capable of self-correction raises ethical and stability concerns, as the technology becomes accessible to a broader range of developers [17]

Group 1: AI Bubble Characteristics - The current AI bubble is characterized as a "rational bubble," differing from historical irrational bubbles like the Dutch tulip mania, due to AI's transformative potential for society[4] - The macroeconomic environment, valuation structures, and investment intensity indicate that the risks associated with the current AI bubble are manageable compared to the 2000 internet bubble[4] - Global major economies are in a rate-cutting window, with actual interest rates significantly declining, which increases market tolerance for long-term capital expenditures and uncertainties in future returns[4] Group 2: Risks and Potential Outcomes - Structural and cyclical issues in the global economy present systemic risks that exceed those during the 2000 internet bubble, driven by demographic aging, debt expansion, and de-globalization[4] - The AI bubble may lead to unexpected liquidity tightening, especially if geopolitical tensions raise commodity prices, potentially resulting in a "secondary inflation" scenario[5] - The difficulty in forming a closed-loop business model within the AI industry could hinder the sustainability of the bubble, as heavy asset expansion increases operational cash flow pressures[5] Group 3: Economic Implications - AI's impact on employment and consumption may lead to economic demand shocks, with the potential for a "smart crisis" that could exacerbate economic downturns[5] - If the bubble bursts, China's adjustment may be more manageable due to its national system and domestic market advantages[5] - The ongoing AI investment is still in the early stages of technology diffusion, with significant room for growth compared to the peak of the internet cycle[7]

大模型市场，直接来了一个「新物种大分叉」。 全球最大AI模型API聚合平台OpenRouter给出数据： 9号到15号这一周，中国大模型的调用量冲到了 4.12万亿Token，第一次超过美国模型的2.94万亿Token。 调用量前五的模型里，中国直接占了四个：MiniMax的M2.5、月之暗面的KimiK2.5、智谱GLM-5、 DeepSeek V3.2。 市场已经彻底分成两拨人：BAT正在「向内收缩」，把模型塞进搜索、电商、办公这些老业务里；新势 力Kimi、智谱、MiniMax则「向外长手」，靠成百上千个Agent死磕开发者生态。 这是商业底层逻辑的彻底分家，大家都在算一笔名为投资回报率的生死账。 01 先得搞清楚一件事：这4.12万亿Token的调用量，到底是谁在用？ 要光看新闻标题，肯定以为中国老百姓突然爱上跟AI唠嗑了；OpenRouter的用户构成很有意 思，上面美国开发者占了47.17%，中国开发者只有6.01%。 这4.12万亿Token，主要是全球那帮程序员「用脚投票」投出来的，跟咱们在抖音刷到的那些AI聊天没 啥关系。 所以，Token调用量比DAU实在多了，日活用户能被红包砸出来，能被 ...