LATIDIA · Ciberseguridad
Dónde residen las pruebas: medición de la privacidad y la diagnosticabilidad de los rastros exportados de los agentes de IA
arXiv:2611.20389v1 Tipo de anuncio: nuevo Resumen: Los registros de ejecución ayudan a diagnosticar los fallos de los agentes de IA, pero compartirlos puede exponer las identidades de los clientes. Por ejemplo, detectar que un agente ha gestionado los registros del cliente equivocado
WhatsApp ↗Telegram ↗
La noticia
arXiv:2611.20389v1 Announce Type: new Abstract: Execution traces help diagnose AI-agent failures, but sharing them can expose customer identities. For example, detecting that an agent handled the wrong customer's records requires knowing whether the records belong to the requested customer, but not necessarily knowing who that customer is. Our executable benchmark measures what a trace export reveals and whether a proposed repair restores the correct output in synthetic multi-tenant workflows. Conventional redaction can remove evidence needed for repair while leaving encoded identifiers recoverable. Two exports preserve ownership comparisons by reporting the match as a comparison fact or by using pseudonyms consistent within each execution. Both preserve the original diagnostic program's repairs.