在网络被关闭的时段，伊朗国营媒体几乎成为唯一的信息源。信息真空之中，大量影像只能在社交平 台上传播，却很难被证实或证伪。 网民和非营利组织通过自发事实核查发现，伊朗官方发布了使用后期编辑和 AI 生成影像，刻意营造 了"反骚乱"的舆论氛围。这类内容在 X 平台上获得了数万次观看。 一场从线下蔓延至线上的舆论战争，正发生在伊朗。 线下，伊朗当局正在组织"反骚乱"集会；线上，断网、媒体管制和信息封锁同时发生。 撰稿 | 陈姚戈、高允毅 编辑 | 王一鹏 与此同时，另一方同样出现大量 AI 生成内容。一段被广泛转发的视频显示，有人从建筑物上扯下国 旗，发布者称有人撤下了伊朗国旗。这段视频经过反向图片搜索后，被证实拍摄于 2025 年 9 月尼泊 尔抗议活动。 非营利组织 WITNESS"技术威胁与机遇"项目副主任 Mahsa Alimardani 指出，在传播过程中被 AI"增 强画质"的现场照片，反而被当局用来否定影像本身的真实性，对抗议事实进行整体抨击。 真假在这一过程中被同时稀释，AI 让"知晓真相"这件事变得更难了。 这并非人们第一次意识到 AI 的风险。近几年，从《要求暂停更强模型训练的公开信》，到《针对 ...

Group 1: Open AI Developments - Open AI has launched a new model, GPT-5.3-Codex, which enhances coding performance and reasoning capabilities, integrating the strengths of previous models and improving speed by 25% [2] Group 2: Xiaomi Automotive Updates - Xiaomi Automotive clarified rumors regarding the dissolution of the SU7 Ultra professional team, stating that sales strategies have been adjusted to improve customer service [4] - The company has reduced the safety mileage threshold for its assisted driving feature from 1000 km to 300 km to help users gradually familiarize themselves with the technology [5] Group 3: Meituan Acquisition - Meituan announced plans to acquire DINGDONG FRESH HOLDING LIMITED for an initial cost of $717 million (approximately 5 billion RMB), with certain conditions regarding cash extraction from the target group [6][7] Group 4: AI and Technology Innovations - Qianwen App launched a "Spring Festival 3 billion free order" campaign, allowing users to receive a 25 RMB no-threshold coupon for use in over 300,000 beverage stores [8] - Xiaoma Zhixing and Moer Thread have formed a strategic partnership to focus on L4 autonomous driving technology, utilizing domestic AI computing power for training and optimization [9] - The International Olympic Committee announced the creation of the first official Olympic large model based on Alibaba's Qianwen, aimed at enhancing service and information delivery during the Milan Winter Olympics [10] - Kuaishou's Keling AI has globally launched the 3.0 series model, which integrates video generation and editing into a unified workflow [11] Group 5: Aerospace and Semiconductor Developments - Fengfei Aviation successfully completed the conversion flight of its 5-ton eVTOL, marking a significant milestone in the eVTOL industry [12] - Intel and SoftBank's SAIMEMORY are collaborating to develop a new type of DRAM storage technology, aiming for prototype release by 2027 and commercialization by 2030 [13] - Texas Instruments announced the acquisition of Silicon Labs for approximately $7.5 billion, aiming to create a leading company in embedded wireless connectivity solutions [14] - Infineon announced a price increase for power switches and related chips starting April 1, 2026, due to rising demand and supply constraints [16] Group 6: Capital Market Activities - Kaide Quartz announced that Jiangfeng Electronics will become the controlling shareholder following a share transfer agreement [17] - Runze Technology is planning to issue convertible bonds to acquire minority stakes in Guangdong Runhui Technology [18] - Guoxuan High-Tech plans to raise no more than 5 billion RMB through a private placement to fund various battery projects [19] - Jinfeng Technology intends to acquire 51% stakes in Zhuohui Metal and Lianyi Thermal for a total consideration of up to 714 million RMB [20]

软银集团：将在日本利用OpenAI的新企业平台开展合资业务服务。 ...

更多精彩内容请关注格林大华期货官方微信 | 板块 | 品种 | 多（空） | 推荐理由 | | --- | --- | --- | --- | | | | | 【行情复盘】 受外盘影响，周四两市主要指数低开下行，下午有所反弹。主要指数呈现破位形态， 叠加纳指趋弱，后市不乐观。两市成交额 2.17 万亿元，继续缩量。沪深 300 指数 收 4670 点，跌 28 点，跌幅-0.60%；上证 50 指数收 3059 点，跌 10 点，跌幅-0.33%； 中证 500 指数收 8146 点，跌 152 点，跌幅-1.84%；中证 1000 指数收 8068 点，跌 | | | | | 139 点，跌幅-1.69%。行业与主题 ETF 中涨幅居前的是银行 ETF 天弘、影视 ETF、 食品 ETF 鹏华、旅游 ETF、消费 ETF 华夏，跌幅居前的是光伏 ETF 华夏、黄金股 ETF 工银、中韩半导体 ETF。两市板块指数中涨幅居前的是影视院线、日用化工、医疗 | | | | | 美容、酒店餐饮、中小银行指数，跌幅居前的是光伏设备、贵金属、工业金属、稀 | | | | IF （弱） | 有金属、风电设备指数。沪深 3 ...

Core Insights - Anthropic has released its latest AI model, Claude Opus 4.6, designed for financial research, which can perform detailed financial analysis that typically takes days for humans to complete [1] - Following the announcement, stock prices of financial service companies plummeted, with FDS down 7%, and declines also seen in SPGI, MCO, and NDAQ [1] Group 1: Company Developments - Anthropic and its competitor OpenAI have invested heavily in AI tools over the past year to simplify professional tasks across various sectors, including financial services and healthcare [2] - Anthropic is negotiating a new funding round at a valuation of $350 billion, while OpenAI's valuation stands at $830 billion [2] - Anthropic has over 300,000 enterprise clients and has become a market leader with its Claude Code model, particularly in simplifying programming tasks [2] Group 2: Market Impact - The release of a legal automation tool by Anthropic has caused significant market turbulence, leading to a drop in stock prices, particularly in the software sector, as investors worry about potential obsolescence [2] - The legal functionality is a plugin for Claude Cowork, which has gained popularity for its intuitive application development and data processing capabilities [3] - Anthropic plans to focus on enhancing its capabilities in cybersecurity, life sciences, healthcare, and financial services [3]

Core Viewpoint - The article discusses the release of OpenAI's latest programming model, GPT-5.3-Codex, which competes with Anthropic's Claude Opus 4.6, highlighting significant improvements in coding capabilities and user interface design [1][13]. Group 1: Model Improvements - GPT-5.3-Codex shows enhanced aesthetic appeal and design in its demos, including a racing game and a diving game [2][6]. - The model has demonstrated superior performance in various benchmarks, achieving 57% in SWE-Bench Pro, 76% in TerminalBench 2.0, and 64% in OSWorld [11][12]. - It has improved efficiency, requiring less than half the tokens compared to its predecessor for the same tasks, with a speed increase of over 25% [11][22]. Group 2: Functional Capabilities - GPT-5.3-Codex excels in computer use, assisting financial professionals in creating presentations and handling complex tasks like document writing and spreadsheet management [9][23]. - The model's ability to self-accelerate during its training process marks a significant advancement, allowing it to monitor and debug its own training tasks [28][29]. Group 3: Business Applications - OpenAI is launching Frontier, a platform aimed at integrating AI into corporate workflows, with notable companies like HP and Uber already adopting it [34][38]. - The AI4S initiative, in collaboration with Ginkgo, aims to reduce protein synthesis costs by 40% using GPT-5, showcasing the model's application in synthetic biology [39][41].

Core Viewpoint - Nvidia may not release a new generation of gaming GPUs in 2026 due to ongoing memory supply shortages, prioritizing memory resources for its AI accelerators instead [1] Group 1: Memory Supply Issues - Nvidia is facing an industry-wide shortage of memory resources, which has led to a reduction in the production of its existing gaming GPU product line [1] - This potential halt in new gaming GPU releases would mark the first interruption in Nvidia's annual product update cycle in thirty years [1] - A company spokesperson acknowledged the memory supply issue, stating that demand for GeForce RTX GPUs is strong while supply remains limited [1] Group 2: Investment in AI - Nvidia is nearing an agreement to invest $20 billion in OpenAI as part of the latter's latest funding round, which would represent the largest single investment by the chipmaker in the ChatGPT developer [1] - This $20 billion investment reflects the intense competition among top tech giants for influence in the AI sector [1]

图片来源： techcrunch OpenAI 正在招聘数百名新员工，以扩充其技术顾问团队，该团队旨在帮助大型企业开发定制化人工智能应用与智能体，实现员工任务的自动化处理。 OpenAI 若能强化自身在大型企业中的信誉，将有助于其在未来的首次公开募股中赢得投资者青睐。据知情人士透露，该公司高管已与投行及律师事务所就 最早于第四季度上市的方案进行非正式磋商。（《华尔街日报》此前曾报道过该公司的预期上市时间表。） 此次大规模招聘或有助于 OpenAI 击退其主要竞争对手 Anthropic 的攻势，后者同样加码了企业服务布局。这一举措正值 OpenAI 准备推出全新企业级解决 方案之际，该方案旨在整合企业应用人工智能的各类需求。 据透露， ChatGPT 的制造商正在扩编其技术顾问——也被称为前线部署工程师的数量，这些工程师能够利用客户自身的资料来定制 OpenAI 的模型。 例 如，这些工程师可以帮助 T-Mobile 开发用于响应客户服务请求的人工智能，或协助 Intuit 为其客户提供报税准备服务。 在约 4,000 名员工中， OpenAI 目前约有 60 名此类工程师，另在与技术紧密相关的客户成功团队中 ...

Investment Rating - The investment rating for the computer industry is "Positive" (maintained) [4][13] Core Insights - The report highlights the emergence of autonomous agents, marking the beginning of the third scaling phase in the industry. Companies like Anthropic are revising their revenue expectations upwards, with projections of $18 billion and $55 billion for 2026 and 2027 respectively. The penetration of agents is expected to drive significant revenue growth [6] - The report notes a decline in global web traffic share for ChatGPT, which fell to 64.5% in January 2026, a decrease of approximately 20 percentage points compared to January 2025. This shift indicates a transition in growth narratives towards mixed revenue models [6] - The report emphasizes the potential for AI-native products to emerge rapidly, amplifying short-term industry volatility while validating long-term trends. It anticipates a significant increase in the "Token->ARR" conversion for agents in 2026, suggesting a robust investment opportunity in high-growth sectors [6] Summary by Sections Section: Market Performance - The report discusses the performance of the computer sector, indicating a positive outlook based on recent developments and market trends [2] Section: Autonomous Agents - The report details the advancements in autonomous agents, including their increasing role in B2B applications and the significant rise in token consumption compared to traditional chat models. The penetration of these agents is seen as a key lever for revenue growth for companies like Anthropic [6] Section: Consumer Productivity - OpenClaw has gained popularity, driving a surge in Mac mini purchases. The report notes that consumer sensitivity to model performance is low, while sensitivity to interaction is high, indicating a shift in user engagement with AI tools [6] Section: Group Intelligence - The emergence of Moltbook, an agent-based platform, is discussed as a preliminary form of group intelligence. The report highlights the initial chaotic nature of agent interactions and the potential for new business models to develop from these early stages [6] Section: Investment Strategy - The report suggests focusing on sectors with clear industry trends and high growth certainty, recommending investments in domestic large models, agents, infrastructure, and computing power companies. It advises on the importance of reverse investment strategies and forward-looking positioning due to industry volatility [6]

Core Insights - The report emphasizes that the recent pullback in the AI industry is a necessary phase in the deep integration of AI technology into various sectors, rather than a fundamental threat to the industry's future [1][6] - It suggests that the current market concerns regarding the uncertainty of AI application business models and hardware demand are part of the industry's evolution, and that this pullback presents investment opportunities in AI applications, cloud services, and storage [2][6] Market Trends - Since mid-January 2026, the AI industry chain has experienced a continuous pullback, exacerbated by several events in early February, including Microsoft's financial report revealing dual concerns about growth dependency and investment returns [2][3] - Microsoft's Q2 2026 financial report indicated a slowdown in Azure cloud computing growth and projected capital expenditures exceeding $100 billion, with approximately 45% of its cloud business backlog dependent on OpenAI [3] - Concerns were also raised regarding NVIDIA's investment stance on OpenAI, with CEO Jensen Huang indicating a cautious approach to investment, despite previous indications of a potential $100 billion investment [4] Business Model Uncertainty - The report identifies dual uncertainties in the market: the uncertainty of AI application business models and the uncertainty of real demand [5][6] - It highlights that traditional SaaS companies may face challenges as enterprises consider building their own AI tools, potentially undermining SaaS profitability [5][6] - The report argues that the market's valuation logic is shifting from paying premiums for future potential to assessing current realities and investment returns [5] Long-term Outlook - The report posits that the concerns regarding business model and demand uncertainties are part of the necessary evolution towards deeper integration of AI technology, rather than a fundamental threat to the industry's prospects [6] - It suggests that traditional application vendors can leverage their industry knowledge and data advantages to build new barriers in the AI era, and that early movers may see valuation increases [6][7] - The demand for hardware is expected to grow in tandem with the maturity of software applications, as AI applications transition from "technology demonstrations" to "production tools" [7]