ISM-2158 · September 2026
Excessive agency
Guidelines for software development / Artificial intelligence application development / Excessive agency
Official control wording
External content retrieved by agentic AI applications is treated as untrusted data throughout processing, is clearly delimited from system instructions, is subject to validation and sanitisation measures applied to other untrusted input, remains untrusted following such processing, and is prevented from modifying or overriding system instructions, security policies, access controls, tool permissions or human approval requirements.
Applicability
These are ASD’s applicability markings. Select and tailor controls for your system’s risk, classification and operating context with the system owner and authorising officer.
Implementation guidance
This control is included as official reference material. We have not published a reviewed Kubernetes implementation mapping for it. Browse the 23 guides with bounded public mappings.
Official source
ASD’s ISM OSCAL v2026.09.4 · Original catalogue JSON · Pinned checksums and edition details
Control source identifier: ism-2158. Check the official release before relying on this snapshot for an assessment.
Official control wording and applicability: Australian Signals Directorate, © Commonwealth of Australia 2026. September 2026 ISM, OSCAL v2026.09.4, reproduced under CC BY 4.0. Formatting is adapted for this website.
Latticework Systems’ implementation guidance is our interpretation. Provenance and scope.