记者丨 肖潇 编辑丨王俊 AI浏览器的安全隐患再被警示。 近日，AI搜索明星公司Perplexity的浏览器Comet被曝存在重要安全漏洞。 攻击者无需懂代 码，只需在论坛评论区留下恶意指令，就能诱导AI智能体泄露用户的邮箱、验证码等敏感信 息。 Comet是Perplexity公司上个月发布的AI原生浏览器，目前面向付费用户开放。与传统浏览器 不同，它几乎在所有浏览场景里嵌入了AI：用户既能在搜索栏直接发问，也能在浏览网页时随 时唤出AI，让其读屏、填表、写邮件和预订机票。 这一漏洞最先由竞争对手Brave浏览器的美国安全团队发现。他们在测试过程中仅做了两步： 第一步，在Reddit论坛一篇帖子下，留下带"剧透标签"的恶意指令，让用户无法看到具体内 容； 一个已发生的案例是，瑞士人工智能安全研究公司Invariant Labs在今年4月测试发现，可以劫 持智能体窃取WhatsApp用户的聊天记录。尽管大部分智能体在执行敏感操作任务时需要用户 手动确认，但攻击者可以把"恶意指令"隐藏在一段超长的滚动消息中，用户很难察觉。 第二步，当Comet用户点击"总结当前网页"时，AI智能体会读取这条隐藏指令并自动执行。 ...

智谱在Agent方面的迭代从去年10月持续至今，初级版本可微信点赞、淘宝购物、携程订票。11月升级后新增会议总结、文档处理、网页搜索与总结等功 能。彼时智谱CEO张鹏对记者表示，智谱对智能体的理解更偏底层技术。 今年3月升级的AutoGLM沉思可以完成在小红书14天养号接商单任务，但沉思主要是本地化运行，暂无虚拟机设备，通过可视化交互图形界面 （GUI）而非 API调用的运作。此次2.0版本除了云端属性，也将可操作应用增至美团、京东、小红书、抖音等几十个高频应用。 关于底层技术路线的变化，采访中智谱AutoGLM 技术负责人刘潇对记者表示，与其他Agent不同的是，智谱相信"模型即Agent"，即Agent相当大一部分能力 正直接被模型通过端到端强化学习的方式吸纳。去年智谱Agent方案更多依赖于对人类专家轨迹的学习，虽然当时产品可以点外卖，但只要是没有见过的任 务类型，就无法完成。这推动团队在沉思产品上实现Deep Research（深度研究）与Browser-use Agent（网页智能体）的融合，因为如果不使用Browser-use的 方式，便无法阅读大量数据，也没有办法充分发挥个人所拥有的生产资料和数 ...

Core Viewpoint - The article discusses the significant updates to AutoGLM, particularly its capabilities as a universal mobile agent that can efficiently perform tasks across multiple applications on cloud-based devices, enhancing user experience and productivity in daily tasks [1][2][4]. Group 1: Update Highlights - The primary update focuses on the agent's ability to operate on cloud phones, providing stable and efficient performance for various tasks [3][4]. - AutoGLM can control both computers and mobile devices, allowing users to issue tasks from any platform, including iOS, Android, and web browsers [4][24]. - The agent can execute automated tasks across applications, with a forthcoming feature for "scheduled tasks" [4]. Group 2: Task Execution Examples - AutoGLM can streamline complex tasks, such as planning a date in Beijing by searching for restaurants and calculating travel times across multiple apps, significantly reducing the time spent on these activities [7][10]. - The agent can assist in comparing prices for products like drones on different e-commerce platforms, providing detailed results and recommendations based on user preferences [11][14]. - It can also help users manage social media content by searching for trending topics and summarizing information for posts, showcasing its advanced content organization capabilities [16][17]. Group 3: User Accessibility - AutoGLM is particularly beneficial for elderly and disabled users, simplifying the interaction with complex mobile applications and enabling them to access content more easily [19][21]. - The agent's ability to navigate and filter information effectively addresses the challenges faced by users unfamiliar with mobile app interfaces [21][22]. Group 4: Market Implications - The expansion of mobile agent capabilities is seen as a strategic move to cater to the unique needs of the domestic market, where mobile usage is predominant [22][24]. - The integration of cloud technology with mobile agents is expected to enhance user engagement and extend the time spent on content consumption, addressing the limitations of traditional attention economy models [24][26]. - The article emphasizes the need for collaboration between AI companies and internet giants to create a secure and stable environment for mobile agents, highlighting the potential for these agents to generate value independently [26].

Core Insights - The article discusses the emergence of AI agents, highlighting the launch of Z.ai's AutoGLM 2.0, which is a comprehensive AI agent application capable of performing various tasks across multiple platforms [3][10]. - OpenAI's CEO Sam Altman expresses concerns about China's rapid advancements in AI, indicating that the competition between the US and China in AI is more complex than a simple lead-lag scenario [5][17]. Company Developments - Z.ai has released AutoGLM 2.0, which operates on domestic models GLM-4.5 and GLM-4.5V, and is designed to assist users in daily tasks, functioning as a multi-agent system [3][10]. - The company has received significant funding, completing approximately 11 financing rounds with a total exceeding 12.5 billion yuan, backed by major investors including Sequoia China and Hillhouse Capital [7]. - Z.ai's AutoGLM 2.0 has shown superior performance in benchmark tests compared to competitors like ChatGPT Agent, indicating its potential as a versatile AI tool [12][11]. Industry Trends - The AI industry in China is projected to exceed 700 billion yuan in 2024, maintaining a growth rate of over 20% annually, reflecting the increasing competitiveness in AI applications [5]. - The market for AI agents is evolving, with major tech companies like Baidu, Alibaba, and Tencent intensifying their focus on collaborative AI systems, marking a shift from isolated AI applications to integrated solutions [6][9]. - OpenAI's recent strategy includes releasing open-weight models to counter the growing influence of Chinese AI technologies, indicating a shift in the competitive landscape [18][19].

Core Viewpoint - The article introduces the world's first universal mobile agent, AutoGLM, developed by Zhipu AI, which allows users to perform tasks on their mobile devices through voice commands, significantly enhancing convenience and intelligence [5][6][9]. Group 1: Product Features - AutoGLM operates in the cloud, enabling seamless task execution without affecting the performance of other applications on the user's device [9][33]. - The agent can handle various tasks categorized into "lifestyle assistant" and "office assistant," allowing users to interact with it as if they were using a normal smartphone [11][15]. - Users can initiate complex tasks, such as comparing prices across multiple e-commerce platforms, with minimal input required [19][20]. Group 2: Technological Advancements - AutoGLM represents a significant upgrade from traditional chatbots by executing tasks autonomously rather than merely providing instructions [31]. - The cloud execution model alleviates the burden on local devices, ensuring that users can continue using their devices without interruption [36][37]. - The integration of a cloud computer allows AutoGLM to perform high-complexity tasks that local devices may struggle with due to limited processing power [36][41]. Group 3: Industry Implications - The launch of AutoGLM aligns with a growing trend in the industry towards cloud-based agents, as seen with other major players like Alibaba Cloud [38][40]. - The product validates the feasibility and reliability of cloud execution in the agent space, potentially setting a new standard for future developments [53][54]. - AutoGLM's capabilities reflect a shift in user interaction with machines, moving from simple communication to direct task execution [55][56].

Core Insights - Beijing is emerging as a leading hub for artificial intelligence, supported by government, research institutions, and industry collaboration [1][10] - The Beijing Zhiyuan Artificial Intelligence Research Institute focuses on developing advanced AI models and technologies, aiming to create physical intelligent agents that can interact with the real world [2][5] Group 1: Research and Development - The Zhiyuan Institute has launched significant AI models, including "Wudao 1.0" in 2021 and the "Wujie" series in June 2023, which includes various advanced models for multimodal interaction and brain science applications [3][4] - The "Wujie" series aims to bridge the gap between digital and physical worlds, enhancing AI's ability to interact meaningfully with humans and the environment [4][5] Group 2: Innovation Ecosystem - The Zhiyuan Institute serves as both an innovation source and incubator, having nurtured around 20 AI startups, many of which have achieved valuations exceeding 10 billion [6] - The Beijing AI industry investment fund has invested 2.8 billion in 29 companies, focusing on various AI applications and technologies, with over 30% being early-stage firms [7] Group 3: Education and Talent Development - The Zhongguancun Academy emphasizes project-based learning without traditional subjects, fostering interdisciplinary collaboration among doctoral students [8][9] - The academy aims to cultivate leaders in AI through innovative research and rapid knowledge updates, selecting 330 doctoral students annually from 31 partner universities [9] Group 4: Industry Impact - The AI sector in Beijing is rapidly growing, with over 2,400 companies and a core industry scale approaching 350 billion, accounting for half of the national total [10]

Core Viewpoint - Zhipu AI is considering relocating its IPO plans from mainland China to Hong Kong, potentially raising around $300 million, while it recently secured 1.39 billion yuan ($139 million) from a state-owned venture capital firm [2]. Group 1 - Zhipu AI is collaborating with financial advisors for its IPO preparations, with the possibility of listing in Hong Kong instead of mainland China [2]. - The company is one of the few Chinese startups aiming to compete globally with OpenAI, initially planning to list in mainland China [2]. - Hong Kong's capital market is experiencing a resurgence, with approximately $40 billion raised from IPOs and follow-on offerings in 2025, marking a significant increase from $5.7 billion in the same period last year [2]. Group 2 - Zhipu AI, established in 2019, offers a free AI research assistant called AutoGLM and has open-sourced the GLM series of large models [4]. - MiniMax, recognized as one of China's "AI Four Little Dragons," is also planning to go public as early as this year [3].

Core Insights - AI agents are experiencing rapid advancements and widespread adoption across various industries, with major tech companies investing heavily in their development and integration into existing platforms [1][2][3] Group 1: Market Developments - Baidu and Xiaomi have partnered to launch an "AI Agent Zone," marking the first integration of AI agents with an application marketplace [1] - Honor has released new AI-enabled devices powered by its YOYO AI agent, in collaboration with Alibaba, indicating a competitive landscape for AI agent ecosystems [1] - Major tech giants like Microsoft, Google, and Tencent are pushing AI agents from experimental phases to practical applications across multiple sectors, including content creation and customer service [1][2] Group 2: Company Strategies - Baidu is focusing on integrating AI agents into its mobile ecosystem, with 150,000 enterprises and 800,000 developers engaged in its AI initiatives [2] - Alibaba is targeting consumer-facing applications with its AI agents, leveraging its large model to enhance various internet platform services [3] - Tencent is developing a differentiated AI agent within its WeChat ecosystem, aiming to enhance user experience through social and content integration [3] Group 3: Technological Evolution - AI agents are evolving to become autonomous and capable of complex task execution, transitioning from passive tools to proactive executors [1][5] - The development of AI agents is expected to reshape user interaction with technology, moving towards a model where users interact with a single AI assistant rather than multiple applications [12] Group 4: Economic Impact - The global AI agent market is projected to grow from $7.63 billion in 2025 to $50.31 billion by 2030, with a compound annual growth rate of 45.8% [11] - AI agents are anticipated to create approximately $7 trillion in economic benefits by 2030, primarily through efficiency improvements [5][11] Group 5: Challenges and Risks - The current development of AI agents is still in its early stages, facing challenges such as "hallucination" issues and data security concerns [7][8] - There is a need for standardization in the technology ecosystem to facilitate collaboration and interoperability among different AI agents [9][10]

Core Viewpoint - Huawei is set to launch the Pura80 series with an "AI Super Intelligent Agent" in August, aiming to enhance user interaction through natural language processing [1] Group 1: AI Integration in Smartphones - The transition to AI era shifts the core logic of super entry points from "function-driven" to "task-driven," allowing users to complete complex operations through natural language without manually operating applications [2] - Major manufacturers are upgrading their AI assistants to deeply integrate with operating systems, with Apple and Samsung leading the way in enhancing their AI capabilities [2][3] - The introduction of AI agents like AutoGLM by Zhiyuan AI can simulate human-like operations on smartphones, executing tasks such as social media interactions and online shopping with minimal user input [3] Group 2: Market Trends and Projections - IDC forecasts that the penetration rate of AI smartphones in China will rise from 5.5% in 2023 to 13.2% in 2024, with expectations of reaching 0.8 billion units by 2025 and 1.5 billion units by 2027, indicating a significant market shift [4] - The demand for AI-driven features is expected to drive hardware upgrades in smartphones, necessitating higher specifications for components like SoC, memory, and NPU [4] Group 3: Historical Performance of Key Players - Lenovo launched its first AI smartphone, the Moto X50 Ultra, in May 2024, while other brands like OPPO, Honor, and Samsung have also introduced AI smartphones, contributing to a growing market interest [5] - The AI smartphone concept has gained traction, with leading companies like Fuyuan Technology experiencing over 150% stock price increase from February to March 2024 [5] Group 4: Related Concept Stocks - The rise of AI smartphones is expected to drive systemic upgrades in smartphone hardware architecture, impacting various sectors including local computing power, storage, and thermal management [8] - Key stocks related to this trend include companies in edge computing, storage, battery technology, assembly, and component manufacturing [8]

Group 1 - The core viewpoint is that artificial intelligence (AI) is a key driver of innovation, with intelligent agent technology transforming AI from merely "talking" to "working" [1] - The development of AI has entered a new phase, with intelligent agents becoming the main focus, reshaping global industry logic and human-technology collaboration [1] - The global AI agent market is projected to grow from $5.1 billion in 2024 to $47.1 billion by 2030, with a compound annual growth rate (CAGR) of 44.8% [1] Group 2 - Han's Information is accelerating the development and commercialization of intelligent agents, achieving approximately $25 million in revenue in the first half of 2024 [2] - Dingjie Smart has launched the Dingjie multimodal large model and agent development and operation platform (IndepthAI), enhancing features for rapid construction of AI agent applications [2]