New watermarking method makes AI text detectable after rewording

Original: Robust Text Watermarking for Large Language Models via Dual Semantic Embeddings

Source: arxiv.org ↗