Guidelines for secure AI system development
These guidelines provide actionable security recommendations for providers of artificial intelligence and machine learning systems throughout the entire system development life cycle. Co-authored by cybersecurity agencies internationally, the document covers secure design, secure development, secure deployment, and secure operation and maintenance. It emphasises a secure-by-design approach, supply chain security, and addressing adversarial machine learning vulnerabilities alongside standard cyber threats.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
This document is aimed primarily at providers and developers of AI systems, as well as data scientists, managers, decision-makers, and risk owners.
What the document contains
The document covers security considerations and mitigations across the four stages of the AI system development life cycle: design, development, deployment, and operation and maintenance. It addresses threat modelling, supply chain security, asset tracking, infrastructure protection, continuous model protection, incident management, and system monitoring.
The document does not provide binding legal regulations, explicit enforcement mechanisms, or formal certification frameworks.
- AI systems are subject to novel security vulnerabilities, such as adversarial machine learning, which must be considered alongside traditional cyber threats.
- Security must be a core requirement throughout the entire AI system life cycle, following a 'secure by default' approach.
- AI component and model providers must take responsibility for the security outcomes of downstream users.
- The guidelines are structured into four key areas: secure design, secure development, secure deployment, and secure operation and maintenance.
· Executive summary · Introduction - Why is AI security different? - Who should read this document? - Who is responsible for developing secure AI? · Guidelines for secure AI system development - 1. Secure design - 2. Secure development - 3. Secure deployment - 4. Secure operation and maintenance · Further reading
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Fairness
1/5Non-discrimination and equitable treatment
Cyber security is a necessary precondition for the safety, resilience, privacy, fairness, efficacy and reliability of AI systems.
Introduction
screening score · 2026-10-23
Human Dignity
0/5Respect for the inherent worth of persons
N/A
screening score · 2026-10-23
Privacy
4/5Protection of personal data and private life
If appropriate, privacy-enhancing technologies (such as differential privacy or homomorphic encryption) can be used to explore or assure levels of risk associated with consumers, users and attackers having access to models and outputs.
3. Secure deployment
screening score · 2026-10-23
Enforcement
0/5Consequences and compliance mechanisms with real force
N/A
screening score · 2026-10-23
Impact Assessment
4/5Required assessment of an AI system's risks/impacts before or during use
As part of your risk management process, you apply a holistic process to assess the threats to your system, which includes understanding the potential impacts to the system, users, organisations, and wider society
1. Secure design
screening score · 2026-10-23