一边是二手市场炒至天价的"一机难求"，一边又是微信、支付宝等大厂的"围剿封锁"；一边是AI代劳的技术畅想，一边又是隐私泄露与合规风险的尖锐拷 问……自亮相以来，豆包手机的争议从未中断。 让各方反应激烈的原因，是豆包触及了手机系统底层的权限，就像AI长出一只"无形的手"，不依赖App和手机厂商，就能模拟人类点击和滑动，且泛用性 极强。 《IT时报》记者将从手机厂商、应用开发者、国外AI手机三方面进行对比，探究豆包手机引发巨大争议的深层逻辑。 12月16日，根据中兴商城官网显示，豆包手机助手及合作机型nubiaM153目前已重启F码候补申请通道，正式恢复产品购买资格。 不同模式同一个目的 《隔屏有手！AI手机自动点咖啡背后，是偷偷启动的无障碍功能》报道半年后，《IT时报》记者再次对各大手机厂商新机进行测试发现，AI助手的智能 化程度得到"史诗级增强"，已经不再局限于点咖啡。 图源：即梦AI "帮我在淘宝上选一条便宜的牛仔裤。"下命令给荣耀新机Magic 8和华为新机Mate 80，不仅可以通过AI助手打开淘宝，搜索"牛仔裤"品类，还能自动比 价，按照指令筛选出"便宜"的商品展现在眼前。如果能给出更详细的尺码和颜色 ...

Core Insights - The article discusses the evolution of mobile AI assistants, highlighting their transition from basic chatbots to advanced personal assistants capable of performing tasks on behalf of users, thus reshaping the AI ecosystem [1][3][4] Group 1: Core Capabilities - Mobile AI assistants are changing the reliance on traditional apps, with major brands like Xiaomi, Honor, Vivo, OPPO, Huawei, and Samsung integrating their own AI assistants into devices [3][4] - Initial capabilities of these AI assistants were overhyped, with real-world success rates for tasks like food delivery being below 3% for most [3][4] - Two main technical routes for mobile AI assistants are identified: intent frameworks that require app cooperation and GUI agents that simulate user actions, with the latter being more prevalent [4][5] Group 2: Privacy and Security - The use of screen-reading capabilities by mobile AI assistants raises significant privacy concerns, as they can access sensitive information like chat logs and banking details [6][7] - The transfer of control to AI assistants poses risks, including potential misinformation and execution errors, which could lead to legal issues [6][7] - Systemic data security risks arise from high-privilege applications operating without external oversight, leading to potential misuse [7][8] Group 3: Commercial Dynamics - The competition between internet apps and mobile AI assistants is intensifying, with concerns that AI could replace human interactions, impacting app engagement metrics and advertising revenues [10][11] - The introduction of AI assistants like Doubao has sparked discussions about the future of app ecosystems and the potential for apps to become mere tools for AI [10][11] - The ongoing struggle for control over user data and the implications of AI's role in transactions highlight the need for clear regulations and responsibilities [12][13] Group 4: Future Considerations - The article emphasizes the necessity for transparent authorization mechanisms and clear accountability in AI operations to establish trust and legitimacy [13][14] - Proposals for giving AI assistants a distinct identity and establishing a regulatory framework are discussed as potential solutions to current challenges [14][15]

Core Viewpoint - The company announced adjustments to the AI operation capabilities of its mobile assistant to ensure stable technology development, industry acceptance, and user experience [2][10]. Group 1: Adjustments to AI Operation Capabilities - The company plans to limit the use of AI in scenarios involving score manipulation and incentive collection to protect the integrity of user interactions [11]. - There will be further restrictions on the use of AI in financial applications, such as banking and online payments, due to concerns over user fund security [11]. - The company will temporarily disable AI capabilities in certain gaming scenarios to maintain fairness in competitive rankings [11]. Group 2: User Feedback and Issues - Users reported issues with logging into WeChat on the company's mobile device, with accounts being flagged for abnormal login environments [6][12]. - The company responded that the ability to operate WeChat through its assistant has been disabled and that affected accounts will be gradually unblocked [12]. Group 3: Compliance and Risks - There are concerns that the AI operations may violate WeChat's service agreement, which prohibits the use of third-party tools for account manipulation [13]. - If deemed as non-official client operations, the company could face restrictions on account functionalities or service denial from Tencent [13]. Group 4: Technology and Permissions - The mobile assistant operates at a deep level within the phone's operating system, requiring extensive software authorizations to control various applications [14]. - The assistant utilizes accessibility permissions combined with AI capabilities, allowing it to simulate user actions and control other applications [14][16]. - The evolution of technology raises risks related to the expansion of permissions and the potential loss of user control over devices [16].

Core Viewpoint - The seminar on "Risks and Governance of Intrusive AI: Dialogue between Law and Technology" highlights the urgent need to address the systemic challenges posed by AI agents, particularly in terms of permissions, data, and accountability, moving beyond theoretical discussions to practical governance solutions [1][2]. Group 1: AI Agent Technical Risks and Safety Mechanisms - The seminar's first session focused on the technical risks associated with AI agents, particularly those utilizing accessibility permissions, which have evolved from assisting individuals with disabilities to becoming autonomous digital assistants capable of executing tasks without user intervention [3][4]. - The expansion of accessibility permissions poses two main risks: the potential for unlimited access to device controls and the blurring of responsibility, as users may lose direct control over their devices [5][6]. - AI agents can operate autonomously, executing complex tasks at speeds far exceeding human capabilities, raising concerns about data privacy and the implications of users granting AI agents access to their data across different applications [5][6]. Group 2: Legal and Ethical Dilemmas - The second session delved into the legal and ethical challenges posed by AI agents, emphasizing the ambiguity surrounding authorization mechanisms and the difficulty in tracing responsibility when actions cannot be recorded [7][8]. - Experts highlighted the need for a clear distinction between the roles of AI agents and traditional users, advocating for the recognition of AI agents as independent entities with their own data pathways to facilitate accountability [7][8]. - The discussion also pointed out the discrepancies in industry standards regarding the use of accessibility permissions, indicating a lack of consensus on the regulatory framework needed to govern AI agents effectively [9][10]. Group 3: Governance Pathways and Industry Practices - The final session explored innovative governance pathways, suggesting a "develop first, regulate later" approach to allow for market growth while addressing compliance risks associated with data ownership and copyright issues [14][15]. - Experts proposed that AI agent liability should not follow a strict no-fault principle but rather a fault-based framework, where service providers must demonstrate due diligence to avoid liability [15][16]. - The seminar concluded with a consensus on the necessity for a differentiated regulatory framework that acknowledges the unique nature of AI agents, emphasizing the importance of collaborative governance involving technology, law, and industry practices [16].

Core Insights - Campus running apps are facing significant backlash from university students due to excessive advertisements and intrusive permissions required for usage [1][2][3] Group 1: App Functionality and User Experience - Campus running apps like Yundong Shijie, Budao Le Pao, and Shandong Campus have become essential tools for tracking students' physical activities, but they have transformed into platforms primarily focused on advertising revenue [2][3][4] - Students are required to download these apps to record their exercise, which has become a mandatory part of their physical education curriculum [3][4] - The apps often require multiple permissions, including accessibility features, which are claimed to be necessary for tracking exercise but are also used to push advertisements [10][11][12] Group 2: Advertising Revenue Model - The revenue model for these apps relies heavily on advertising clicks and user engagement, with each interaction generating income for the app developers [13][16] - The apps have been reported to use at least 20 different advertising SDKs, indicating a robust advertising strategy that capitalizes on user attention [11][12] - The cost for universities to implement these apps is relatively low, with contracts often ranging from 1.6 million to 7 million yuan, suggesting a high-profit margin primarily derived from student engagement rather than direct service costs [15][16] Group 3: Privacy Concerns and Regulatory Issues - The use of high-sensitivity permissions, particularly accessibility features, raises significant privacy concerns as they allow the apps to perform actions without user consent [12][20] - Legal experts have pointed out that the bundling of permissions and the forced consent for non-essential data collection may violate privacy laws [19][20] - There is an ongoing debate about the ethical implications of using accessibility features for commercial purposes, with potential regulatory scrutiny on the practices of these campus running apps [19][20]

Core Viewpoint - Campus running apps are facing significant backlash from university students due to excessive advertisements and intrusive permission requests, transforming from tools for sports management into platforms for advertising revenue generation [1][4][10]. Group 1: App Functionality and User Experience - Campus running apps like Yundong Shijie, Budao Le Pao, and Shandong Campus have been adopted by over 700 universities in China, but they are criticized for their overwhelming advertisements and poor user experience [1][4]. - The apps require extensive permissions, including accessibility features, which are claimed to be necessary for tracking exercise but are also used to push advertisements [10][11]. - Users report that the apps often redirect them to shopping platforms and bombard them with ads, leading to frustration and negative feedback [1][6][8]. Group 2: Financial Aspects and Market Dynamics - The cost for universities to install the Yundong Shijie app is approximately 2.88 yuan per student per year, indicating a low-cost model for schools but a high potential for revenue generation through student engagement [2][14]. - The revenue model for these apps relies heavily on advertising clicks and user engagement, with a significant number of third-party ad SDKs integrated into the apps [11][13]. - Despite the low development costs for these apps, the real profit lies in the attention and data collected from students, as they are often required to use these apps regularly [16][17]. Group 3: Privacy and Ethical Concerns - The use of accessibility permissions in these apps raises ethical questions, as they can track user behavior and automate actions without explicit consent [10][22]. - Legal experts argue that the forced consent for multiple permissions violates privacy laws, as users should not be compelled to agree to unnecessary data collection [21][22]. - The apps' practices of using high-risk permissions for advertising purposes have led to scrutiny and potential regulatory actions against them [22].

Core Viewpoint - The article discusses the dual nature of "accessibility permissions" in AI smartphones, highlighting their essential role in aiding individuals with disabilities while also posing significant privacy risks due to potential misuse by malicious actors [1][2][8]. Group 1: Accessibility Features and AI Integration - Accessibility features, originally designed for individuals with disabilities, are becoming mainstream with the rise of AI smartphones, allowing broader user interaction [1][4]. - Major smartphone manufacturers have integrated these features into their AI assistants, enabling tasks such as ordering coffee or sending messages through voice commands [2][5]. - The implementation of accessibility functions is driven by regulatory requirements, such as China's "Accessibility Environment Construction Law" and the EU's "Accessibility Act," which mandate compliance for digital platforms [4][5]. Group 2: Privacy Risks and Concerns - The use of accessibility permissions allows AI assistants to read all screen content, including sensitive information like banking details and passwords, raising significant privacy concerns [5][8]. - There have been numerous cases of abuse where accessibility features were exploited for malicious purposes, such as unauthorized data collection and financial fraud [9][10]. - The potential for continuous data monitoring and user profiling through AI assistants poses a serious risk, as these systems can gather extensive personal information without user awareness [8][12]. Group 3: Regulatory and User Awareness Challenges - Users often lack a clear understanding of how their data is used and the risks associated with granting permissions to AI assistants, leading to unintentional privacy breaches [12][13]. - Current privacy policies and permission requests are often vague, making it difficult for users to make informed decisions about their data [12][13]. - Experts emphasize the need for better user education on privacy management and the importance of being vigilant about app permissions [14][15].