中文

机器人可窥探:揭示群聊中机器人隐私风险及缓解措施

密码学与安全 2025-02-07 v2

摘要

随着群聊平台上的聊天机器人,新的隐私问题不断出现。聊天机器人可能访问其意图功能之外的信息,例如发件人身份或机器人无意访问的消息。聊天机器人开发者可能利用此类信息推断个人信息并将用户在不同群组中关联起来, potentially 导致数据泄露、持续跟踪或定向广告。在我们分析对话数据集后显示, (1) 聊天机器人 often 访问远超所需的消息数量, and (2) 当用户加入包含聊天机器人的新群组时, 至少有一个聊天机器人识别并将用户与其在其他群组中的以前交互关联起来的概率为3.6%。 Although 目前最先进(SoA)的群聊协议提供了强大的端到端加密且一些平台已实施限制聊天机器人访问的政策, 但没有任何平台成功地将这些功能结合在一起。 This paper introduces SnoopGuard, a secure group messaging protocol that ensures user privacy against chatbots while maintaining strong end-to-end security. Our protocol offers (1) selective message access, preventing chatbots from accessing unrelated messages, and (2) sender anonymity, hiding user identities from chatbots. SnoopGuard achieves O(logn+m)O(\log n + m) message-sending complexity for a group of nn users and mm chatbots, compared to O(log(n+m))O(\log(n + m)) in SoA protocols, with acceptable overhead for enhanced privacy. Our prototype implementation shows that sending a message to a group of 50 users and 10 chatbots takes about 10 milliseconds when integrated with Message Layer Security (MLS).

引用

@article{arxiv.2410.06587,
  title  = {Bots can Snoop: Uncovering and Mitigating Privacy Risks of Bots in Group Chats},
  author = {Kai-Hsiang Chou and Yi-Min Lin and Yi-An Wang and Jonathan Weiping Li and Tiffany Hyun-Jin Kim and Hsu-Chun Hsiao},
  journal= {arXiv preprint arXiv:2410.06587},
  year   = {2025}
}

备注

20 pages, 8 figures, In Proceedings of the 34rd USENIX Security Symposium