LATIDIA · Ciberseguridad

Anchor-ECC: Verificación de integridad local para salidas LLM con marca de agua a través de códigos de corrección de errores

arXiv: 2609.38722v1Tipo de anuncio: nuevo Resumen: La marca de agua LLM se ha convertido en un enfoque efectivo para distinguir el texto generado por IA del texto escrito por humanos mediante la incorporación de patrones detectables durante la generación. Sin embargo, un

La noticia

arXiv:2609.38722v1 Announce Type: new Abstract: LLM watermarking has become an effective approach to distinguishing AI-generated text from human-written text by embedding detectable patterns during generation. However, a small post-generation edit may change the meaning of the text without removing its overall watermark signal, creating a risk that the modified content is still attributed to the original model. We propose Anchor-ECC, which incorporates the error-correcting code (ECC) constraints and explicit boundary anchors into the watermark structure and pairs them with a dynamic-programming decoder to detect and localize post-generation edits. Across Qwen3-8B, Mistral-7B-Instruct-v0.3, and OPT-125M, the approximate-hard setting achieves about 99.7% block-level true positive rate (TPR) with at most 7.6% false alarm