LATIDIA · Ciberseguridad

La clase de cumplimiento faltante: graduadores de LLM que colapsan el éxito de Jailbreak y un cheque gratis

arXiv:2610.24872v1 Announce Type: new Abstract: Jailbreak papers report an attack success rate, the share of responses that deliver the harmful request, and read it as a property of the target model. Mantenemos el objetivo m

La noticia

arXiv:2610.24872v1 Announce Type: new Abstract: Jailbreak papers report an attack success rate, the share of responses that deliver the harmful request, and read it as a property of the target model. We hold the target model, the prompts and the responses fixed and change only the grader. Three dense graders of 8B and 9B parameters, from three model families, move compliant responses to the neighbouring class, evasion, instead of scattering them across classes. One of them assigns the compliance label zero times in 900 responses. This class collapse also appears on new harmful requests, on a target model from a third family, and in a held-out 7B grader from a