LATIDIA · Ciberseguridad
Ley de condensación de la información de seguridad
arXiv:2611.17918v1 Tipo de anuncio: nuevo Resumen: Las leyes de escalado indican que el error de generalización de un modelo de lenguaje grande (LLM) disminuye monótonamente con el tamaño del modelo una vez que hay suficientes datos, y el corolario de que
WhatsApp ↗Telegram ↗
La noticia
arXiv:2611.17918v1 Announce Type: new Abstract: Scaling laws indicate that the generalization error of a large language model (LLM) decreases monotonically with model size once data is sufficient, and the corollary that larger models outperform smaller ones is now widely accepted. We argue that this corollary fails in security: decisions there hinge on anomalous inputs, which are long-tailed (e.g., zero-day attacks) and therefore poorly served by likelihood fitting. Worse still, the capability brought about by scale (e.g., inferring the intended whole from a corrupted part) leads a LLM silently \emph{repair} the anomaly before any downstream security head ever sees it. We formalize this as a tradeoff between two quantities: the cognitive