LATIDIA · Ciberseguridad

Caracterización de firmas recuperables de utilización de expertos en modelos LLM de mezcla de expertos finamente ajustados

arXiv:2610.41039v1 Anuncio Tipo: cruzado Resumen: A medida que los grandes modelos de lenguaje Mixture-of-Experts (MoE) se implementan cada vez más en entornos distribuidos, desde el aprendizaje federado hasta los sistemas edge-cloud, su arquitectura

La noticia

arXiv:2610.41039v1 Announce Type: cross Abstract: As Mixture-of-Experts (MoE) large language models are increasingly deployed in distributed environments, from federated learning to edge-cloud systems, their architecture harbors an underexplored landscape of security vulnerabilities. Moving beyond textual analysis, we uncover structural memorization: a novel vulnerability where a fine-tuned model memorizes not only its training data but also the internal pathway used to process it, creating a characteristic routing fingerprint unique to the training process. We provide comprehensive empirical evidence that structural memorization emerges consistently even from a few samples, and that its intensity is directly influenced by the choice of trainable modules. We show that these memorized expert pathways can be stealthily