iSeal: Encrypted Fingerprinting for Reliable LLM Ownership Verification
Published in AAAI Conference on Artificial Intelligence (AAAI '26), 2025
iSeal is the first encrypted fingerprinting method designed for reliable LLM ownership verification when the model thief fully controls the LLM inference pipeline. The full-length version is available on arXiv.
Recommended citation: Zixun Xiong, Hao Wang, Qingyang Yu, Gaoyi Wu, et al. 2025. "iSeal: Encrypted Fingerprinting for Reliable LLM Ownership Verification." In Proceedings of the AAAI Conference on Artificial Intelligence (AAAI '26).
Download Paper