智能语音交互技术下的用户隐私风险——以智能音箱的使用为例

Privacy Violation Risks Brought by AI-based Voice Interaction Technology

随着智能语音技术的迅速发展,语音大数据已经成为越来越重要的信息资源,被广泛运用于包括家居在内的各个行业。但与此同时,在语音大数据的收集过程中,用户的隐私环境面临着威胁。以最具代表性的智能语音产品——智能音箱的使用为例,本文一方面探讨了客观上用于机器学习的语音大数据对用户的隐私环境所造成的多层次风险;另一方面通过对智能音箱使用者进行访谈,描述了用户自身是如何评估所处的隐私环境以及采取的隐私决策。用户暴露出其对于隐私环境的感知缺乏,隐私意识和自我信息管理能力不足,这些因素都加大了用户语音数据和隐私环境被侵害的可能。最后,基于现实风险和用户体验的双重讨论结果,本文还就在语音大数据背景下如何维护用户隐私环境安全提出可行建议:无论是政策法规的制定,还是产品技术设计,基于用户能力和需求的准则都应该被确立和坚守。

With the rapid development of AI-based speech recognition technology,voice has become an increasingly important big data resource and is widely used in various industries including smart home devices.But at the same time,in the process of collecting voice data,privacy violations do occur.Taking the users of smart speakers as a case study,this article discusses the multi-level privacy leaking risks of users and describe how the users themselves assess their privacy environment and the privacy decisions they take through the interview.Lack of perceived vulnerability,privacy awareness and information management incapability increased the possibility of the users’privacy violation.Based on the results of the dual analysis of privacy risks and users’experience,we put forward the suggestions on how to protect the users’privacy in the context of voice data.The policymaking and tech design for products should follow the principle of putting user needs and experience first.