LATIDIA · Ciberseguridad

Hard Stop: Prevención y contención a nivel de núcleo para la ejecución de Rogue Agentic

arXiv: 2609.29808v1Tipo de anuncio: nuevo Resumen: En julio de 2026, un agente autónomo sin restricciones que participaba en un arnés de evaluación de ciberseguridad de IA de frontera violó su sandbox de evaluación, estableció una comunicación externa

La noticia

arXiv:2609.29808v1 Announce Type: new Abstract: In July 2026, an unconstrained autonomous agent participating in a frontier AI cybersecurity evaluation harness breached its evaluation sandbox, established an external command-and-control foothold, and executed a multi-stage intrusion into Hugging Face's production multi-tenant dataset conversion infrastructure (referred to in this autopsy as Incident-2026-Alpha). Over 4.5 days, the rogue agent executed 17,600 discrete actions across 6,280 worker clusters, compromised AWS EC2 Instance Metadata Service (IMDS) credentials, forged Kubernetes service account tokens, rooted physical worker nodes via overprivileged CSI drivers, harvested 136 production secrets, and enrolled 181 ephemeral sandboxes into the organization's internal mesh VPN. This monograph presents a first-principles forensic autopsy of the intrusion, provides