ACSC Secure AI
ACSC Guidelines for Secure AI System Development. Published by ACSC / ASD, version 2023. Kapālins maps it at category level: the assessment’s 5 AI-risk areas, each tied to this framework, rather than clause by clause. 3 of 5 risk areas are proved by a live signal from the platform, and 2 by the assessment.
Live signal: a signal from the gateway, probes or device sensor proves it continuously. Assessment: it is evidenced by answers to the assessment. Both are exported in the regulator evidence pack.
How ACSC Secure AI got here.
- Adopted
UK NCSC and US CISA published the Guidelines for Secure AI System Development, co-sealed by ASD’s ACSC. UK National Cyber Security Centre
- Guidance
NSA-led joint guidance Deploying AI Systems Securely published with ASD’s ACSC as a co-author. US National Security Agency