Get Started
Home
Topics
Search
Library
Research questionHow can watermarking preserve reliable provenance detection after LLM text is paraphrased or translated?Paraphrasing and translation can preserve a text’s meaning while weakening the signals used to identify its origin. This makes provenance verification unreliable for transformed LLM-generated text.
AI
Natural Language Processing
Research Paper
Technology
Latest papersRecent research connected to this question, newest first.Robust Text Watermarking for Large Language Models via Dual Semantic EmbeddingsThe source concerns watermarking based on contextual and token-level embeddings, with watermark signals obfuscated using secret-key projections. Its experiments address robustness to paraphrasing and translation, text quality at lower watermark strength, and computational overhead; the supplied summary does not specify deployment or detector-access conditions.research paper · Sep 4, 2026
Related questions
How can watermarking provide trustworthy provenance for LLM text at scale despite transformations and accumulating false positives?How can LLM watermarks reduce vocabulary-membership overhead without weakening statistical detection guarantees?How can watermarks in generated speech remain detectable after neural audio transformations?How can authorship attribution remain reliable for long-form LLM text across languages and distribution shifts?