LATIDIA · Ciberseguridad

A Deterministic Security Runtime for LLM Agents on MPI and Batch-Scheduled HPC Systems: Consensus-Governed, Hardware-Rooted Supervision of Every Rank

arXiv: 2610.33318v1Tipo de anuncio: nuevo Resumen: Los agentes de modelos de lenguaje grandes lanzan simulaciones en sistemas de computación de alto rendimiento (HPC), donde un sbatch o mpirun puede comprometer miles de horas de nodo y un resultado falsificado

La noticia

arXiv:2610.33318v1 Announce Type: new Abstract: Large language model agents launch simulations on high-performance computing (HPC) systems, where one sbatch or mpirun can commit thousands of node-hours and a forged result can cost lives. Of 23 surveyed systems, none supervises the running job, and none holds when the control host is compromised. We fix both. Enclawed, our deterministic security runtime for AI agents, has until now served a single user; this work makes it a distributed runtime for parallel HPC and agent swarms. The Enclaweder, our hardware root of trust, keeps a compromised host from becoming a catastrophe. A launch gate decides every launch by policy alone, refusing what it cannot