SoK study: agentic jailbreaks hit planning and tools while safety dashboards stay green
An SoK paper submitted to arXiv on September 11, 2026 by Mia, Wu, Uluagac and Amini argues attacks can succeed at an agent’s planning, memory or tool layers while the final response looks clean. TechTimes placed it on September 15 beside Amodei’s swarm warnings; it urges intermediate logging, least privilege and stopping reliance on output-only metrics.