豆包手机遭“五大派”围剿，实测23款主流App，哪些“使不动”？

Core Viewpoint - The launch of Doubao mobile assistant has sparked significant attention in the tech industry, with discussions surrounding its functionality, security, and potential to disrupt existing internet business models [3][27]. Group 1: App Functionality and Limitations - Out of 23 tested apps, 8 faced usage restrictions, while 15 remained fully operational with AI capabilities [4][28]. - Notable apps that could not be logged into include Taobao, Meituan, Alipay, Pinduoduo, and Maoyan, while apps like Didi, QQ, and JD remained unaffected [4][28]. - The restrictions primarily affected apps from major companies such as Tencent, Alibaba, Meituan, and Pinduoduo, indicating a coordinated response to the Doubao assistant's capabilities [5][30]. Group 2: User Experience and AI Assistant Capabilities - The Doubao mobile assistant demonstrated practical AI capabilities, such as effectively closing ads and navigating app interfaces without user intervention [31][33]. - It showcased flexibility by attempting multiple paths to complete tasks, such as navigating to different map applications when faced with restrictions [11][37]. - The assistant could also remember user preferences and make recommendations based on previous interactions, enhancing the user experience [40][42]. Group 3: Technical Aspects and Permissions - The smooth operation of the Doubao assistant is attributed to its INJECT_EVENTS permission, allowing it to execute tasks in the background without user awareness [18][44]. - The assistant's functionality is built on the UI-TARS model, optimized for mobile use, which enhances its ability to interact with various applications [19][45]. - Concerns were raised regarding the potential for the assistant to breach app security measures, as it operates with system-level permissions that could be perceived as intrusive [20][46]. Group 4: Challenges and Industry Implications - The Doubao assistant faces challenges related to privacy concerns and the disruption of existing business models, as it could undermine the revenue streams of traditional apps reliant on user engagement [22][48]. - The potential for resistance from app developers is significant, as they may implement measures to block the assistant's access to their services [22][48]. - Experts suggest that forming a cooperative ecosystem among app developers may be necessary to address the conflicts arising from the introduction of AI assistants [22][49].