BOLT: Privacy-Preserving, Accurate and Efficient Inference for Transformers
Qi Pang, Jinhao Zhu, Helen Möllering, Wenting Zheng, Thomas Schneider
IEEE Symposium on Security and Privacy 2024 · Day 3 · Continental Ballroom 6
The proliferation of powerful Transformer-based models, such as GPT, BERT, and ViT, has fueled the rapid expansion of Machine Learning as a Service (MLaaS). While these models offer unprecedented performance in tasks ranging from chatbots to translation, their deployment raises significant privacy concerns. Incidents like ChatGPT's temporary ban in Italy due to user data exposure highlight the critical need for robust privacy protections in MLaaS. Users' sensitive inputs and chat histories, if revealed in plaintext, can leak personal identities, hobbies, and even commercial secrets, posing substantial risks to individual privacy and corporate data security.
AI review
This work on BOLT is a critical advancement for privacy-preserving MLaaS, addressing a fundamental trust problem with concrete, measurable improvements. The co-design of cryptographic primitives and ML optimizations delivers a truly practical system for secure Transformer inference, pushing the state of the art significantly. This isn't just theory; it's a blueprint for deployable confidential AI.