南财引领智能体健康发展 行业协会发标准：禁止滥用无障碍服务

Core Insights - The year 2025 is anticipated to be the inaugural year for AI agents, with rapid applications leading to risks such as data breaches and misuse of accessibility permissions [1][3] Group 1: Industry Developments - The Nanfang Compliance Technology Research Institute and the 21st Century Economic Report have initiated a series of reports starting December 2024, testing various AI mobile agents and identifying unclear responsibilities regarding data security among participants [1][4] - The China Software Industry Association has released a group standard titled "Accessibility Security Requirements for Mobile Internet Services," addressing issues raised in previous reports, including the prohibition of misuse of accessibility services and the protection of user rights [2][7] Group 2: Risks Associated with AI Agents - AI-driven applications, such as smart assistants, can understand user intent and automate tasks, but they also pose data security risks, including potential data leaks and privacy violations due to the misuse of accessibility services [3][4] - The reliance on cloud-based large model architectures for AI agents raises concerns about data privacy, as users are often unaware of how their data is collected, processed, and stored [5] Group 3: Recommendations and Standards - The Nanfang Compliance Technology Research Institute has proposed initiatives to enhance the management of accessibility permissions and improve transparency in their usage [6] - The group standard emphasizes the need for explicit user consent before enabling accessibility services, the provision of clear information regarding data collection and processing, and the principle of "dual authorization" for executing tasks through third-party apps [7]