LATIDIA · Ciberseguridad

SecPO: Entrenamiento adversario basado en principios para la seguridad de inyección rápida

arXiv:2610.41998v1 Anuncio Tipo: nuevo Resumen: Reconocida como una de las principales amenazas para los agentes de IA, la inyección de mensajes permite que las instrucciones de datos no confiables (correo electrónico, documentos, sitios web) manipulen el funcionamiento del agente. La exis

La noticia

arXiv:2610.41998v1 Announce Type: new Abstract: Recognized as a top threat to AI agents, prompt injection allows instructions from untrusted data (email, documents, websites) to manipulate the agent's operation. The existing best-defended LLM, Meta-SecAlign, is trained with the DPO loss but still suffers from attack success rates (ASRs) approaching 100% under adaptive prompt injections. One reason is that DPO anchors the model around the undefended reference model, which is insecure. We show an idealized reference model---an injection-invariant, utility-preserving oracle---can nevertheless be approximated, as its output distribution in training is interestingly accessible. This new observation motivates a theoretically principled SecPO loss. Another Meta-SecAlign shortcoming is that it trains on weak simulated attacks.