Securing Agentic AI - Addendum to the Guidelines and Companion Guide on Securing AI Systems
Is this legally binding?
Guidance. Voluntary guidance. Best practice, not obligation, until a contract or a regulator cites it.
CSA's final Addendum on securing agentic AI systems, officially published 17 June 2026 (version 1.0) after public consultation 22 October - 31 December 2025. Addresses rogue actions and sensitive data disclosure via autonomy assessment, threat modelling and lifecycle controls. Voluntary.
From the source
“This document is intended for informational purposes only and is not mandatory, prescriptive nor exhaustive.”
Addendum v1.0 (June 2026), Version History
Instrument record
- Short name
- CSA Agentic Addendum
- Date
- 2026-06-17
- Version
- v1.0
- Cluster
- core
What this connects to
6 relations. Official relations are the ones the source documents state; anything marked GAGE analysis is our reading, not an agency's.
Cites1
- InstrumentCSA Guidelines and Companion Guide on Securing AI SystemsGuidance
Addendum is designed to be read alongside, and builds on, the 2024 Guidelines and Companion Guide
Cited by2
- InstrumentModel AI Governance Framework for Agentic AIGuidance
Agentic MGF footnotes 1, 12 and text adapt from CSA Draft Addendum on Securing Agentic AI
- ObligationPerform threat modelling and taint tracingGuidance
'organisations may refer to CSA's Draft Addendum on Securing Agentic AI' for threat modelling and taint tracing
Maps across to2
- External frameworkArticle 15: Accuracy, robustness and cybersecurityBindingGAGE analysis, not official
CSA's agentic security controls directly address Article 15 cybersecurity and robustness concerns
- External frameworkClause 6: PlanningStandardGAGE analysis, not official
Agentic autonomy-aware risk assessment aligns with AIMS risk planning
Verified against the official source on 2026-08-17. GAGE is not affiliated with or endorsed by any agency named here, and nothing on this page is legal advice. How this is built and checked.