LATIDIA · Ciberseguridad

PerMIT: Asegurar la memoria persistente para los agentes de LLM a través de teoremas de integridad verificados por máquina y confianza comunitaria escalable

arXiv: 2610.28987v1Announce Type: cross Resumen: La memoria persistente es una superficie en gran medida indefensa para la inyección y el envenenamiento, y el mismo riesgo limita el intercambio de memoria entre los agentes. Presentamos PerMIT, un arquitecto de memoria

La noticia

arXiv:2610.28987v1 Announce Type: cross Abstract: Persistent memory is a largely undefended surface for injection and poisoning, and the same risk limits memory sharing among agents. We present PerMIT, a memory architecture with provable integrity guarantees under an explicit trust assumption: only user-typed inputs and values confirmed at a gate are trusted. PerMIT mediates the release and use of untrusted environment-derived values through a finite memory interface, while preserving agent-driven multi-turn planning and tool choice. Over this finite interface, we prove two integrity theorems that are also machine-checked in Lean 4, which together ensure that environment-derived memory reaches a sensitive action only if confirmed at the gate and influences the agent's