LATIDIA · Ciberseguridad

LineageMark: Marca de agua de caja blanca multiusuario para el seguimiento de contribuciones en cadenas de derivación de modelos

arXiv:2606.17123v2 Anuncio Tipo: reemplazar Resumen: En los ecosistemas abiertos de modelos de lenguaje grandes (LLM), los modelos se adaptan frecuentemente a través de múltiples dominios y aplicaciones, formando cadenas de derivación de múltiples etapas. Consecuente

La noticia

arXiv:2606.17123v2 Announce Type: replace Abstract: In open large language model (LLM) ecosystems, models are frequently adapted across multiple domains and applications, forming multi-stage derivation chains. Consequently, tracking and verifying historical contributions is essential for model provenance and intellectual property protection. However, existing watermarking methods are mainly designed for single-user, one-time embeddings, often fail under repeated model derivation and incremental updates. To address this problem, we propose LineageMark, a multi-user white-box watermarking framework for model derivation chains. The framework encodes watermarks in model parameters using a projection-based approach. Stable carriers are first selected to reduce sensitivity to model changes, each watermark bit is then represented as a projection statistic over these