Pseudorandom Error-Correcting Codes Enable Provable AI Watermarking
This research introduces Pseudorandom Error-Correcting Codes (PRCs), a novel cryptographic primitive providing provable guarantees for watermarking generative AI models.
