LATIDIA · Ciberseguridad

Estegomalware basado en permutaciones en modelos de lenguaje grandes: amenazas y contramedidas

arXiv:2609.16193v1 Tipo de anuncio: nuevo Resumen: La dificultad de entrenar modelos de lenguaje grandes (LLM), junto con su ubicuidad, plantea la amenaza de stegomalware, donde las cargas útiles maliciosas están incrustadas en el modelo w

La noticia

arXiv:2609.16193v1 Announce Type: new Abstract: The difficulty of training large language models (LLMs), together with their ubiquity, raises the threat of stegomalware, where malicious payloads are embedded into model weights. Recent work has demonstrated the use of permutation symmetry in model weights to mitigate these threats, but failed to show neutralization of stegomalware across all weights for LLMs. In this paper, we demonstrate the full potential of behavior-preserving symmetries as a defense against stegomalware, as well as the risks these symmetries pose when exploited by attackers. For stegomalware neutralization, we improve upon previous work, demonstrating that it is possible to select permutations which displace all model parameters. This contrasts with