Abstract: With the widespread adoption of artificial intelligence (AI) technology, ensuring the privacy of user data has become a significant research concern. Enterprises establish privacy policies to safeguard user data privacy. However, most of these policies exhibit several weaknesses, such as lack of transparency, opacity, unfairness, low practicality, and timeliness. Furthermore, they often overlook predictive private information provided by users. Consequently, these privacy policies are ineffective in protecting acquired privacy and may inadvertently violate predictive privacy, contradictory to their original intent, creating a privacy policy paradox. Drawing upon the human-centric AI framework, we suggest the development of Human-Centered privacy policies to address the challenges posed by current privacy policies effectively and ensure the data privacy security of users.

摘要： 随着人工智能技术的广泛应用，用户的数据隐私安全成为研究的焦点之一。为了保护用户的数据隐私安全，企业提供了隐私政策。然而，大多数隐私政策不仅存在不可解释、不透明、不公平、实用性不强和及时性较低等问题，还忽略了用户的预测隐私。因而在具体实践中，隐私政策既不能有效保护直接获取得到的隐私，又会侵犯到预测隐私，违背了最初的目的，产生了隐私政策的保护悖论。本文参照以人为本的人工智能框架，提出要开发以人为本的隐私政策，以尽可能消解隐私政策的困境，确保用户的数据隐私安全。