LATIDIA · Ciberseguridad

No es autodecidible: los LLM no pueden trazar el límite de lo que un agente verificador puede comprobar.

arXiv:2610.04699v1 Tipo de Anuncio: Cross Resumen: Un verificador para un agente se enfrenta a reglas de dos tipos: las que un cheque fijo puede resolver y las que requieren un juez. Un equipo que deriva sus propias correcciones de cheques que se dividen

La noticia

arXiv:2610.04699v1 Announce Type: cross Abstract: A verifier for an agent faces rules of two kinds: the ones a fixed check can settle and the ones that require a judge. A team that derives its own checks fixes that split up front. Where the requirements come from outside, as in finance, healthcare and law, the agent enforces rules it did not write, so the split falls to runtime, recurring for every predicate of every rule on every action at a rate no reviewer can audit. Every escalation scheme assumes a model can make that decision itself, that it is self-decidable. Across six corpora, including the EU AI Act, FINRA guidance and