Free course · 10 lessons · quizzes · certificate

ISO/IEC 27090 Explained: Securing AI Systems

ISO/IEC 27090:2026 is the international standard on security threats and compromises to AI systems, published by ISO/IEC JTC 1/SC 27 on 14 October 2026. This free course explains the standard in ten lessons: what it covers, the twelve threats it catalogues, how to detect and mitigate them, and how to implement it within an ISO/IEC 27001 programme.

About 3 hours including the exam 50 quiz questions plus a 20-question exam No account needed

About the standard

Standard
ISO/IEC 27090:2026
Full title
Cybersecurity — Artificial intelligence (AI) — Addressing security threats and compromises to AI systems
Published
14 October 2026, edition 1
Committee
ISO/IEC JTC 1/SC 27 (information security, cybersecurity and privacy protection)
Length
56 pages, English
Type
Guidance (informative). No certification scheme.
Normative references
ISO/IEC 22989:2022 (AI concepts and terminology), ISO/IEC 23053:2022 (framework for AI systems using machine learning)

What you will learn

The course follows the standard from first page to last: what it is and is not, why AI systems need their own security guidance, the life cycle and roles, the standards it connects to, the twelve threats it catalogues with their detection and mitigation, the mitigation principles in clause 7, and a practical implementation path with templates. Each lesson ends with a five-question quiz. The final examination draws twenty questions from the whole course and issues a certificate at 80 percent or higher.

01

State the scope, status and structure of ISO/IEC 27090:2026

02

Apply confidentiality, integrity and availability to AI assets

03

Map threats to the seven life-cycle stages and assign responsibility

04

Describe all twelve threats with consequences, detection and mitigation

05

Explain why controls interact and how to keep them effective over time

06

Implement the standard inside an ISO/IEC 27001 programme with templates and a roadmap

The 10 lessons

0 of 10 lessons complete0%
  1. Lesson 112 min

    What is ISO/IEC 27090:2026?

    ISO/IEC 27090:2026 explained: what the AI security standard covers, who it applies to, what it is not, and why it was written. Lesson 1 of the free course.

    Open lesson
  2. Lesson 212 min

    Why AI systems need their own security guidance

    Why AI systems create new attack surfaces, which assets ISO/IEC 27090 protects, and how confidentiality, integrity and availability apply to them.

    Open lesson
  3. Lesson 311 min

    The AI life cycle and who is responsible

    The seven AI life-cycle stages, where each threat enters, and how ISO/IEC 27090 divides responsibility between providers, deployers and platforms.

    Open lesson
  4. Lesson 413 min

    Where ISO/IEC 27090 fits: standards, frameworks and regulation

    How ISO/IEC 27090 relates to ISO/IEC 27001, 27002, 42001 and 27091, how it is organised, and how it maps to the EU AI Act, NIST, OWASP and MITRE ATLAS.

    Open lesson
  5. Lesson 511 min

    The threat catalogue: how ISO/IEC 27090 classifies threats

    How clause 6 of ISO/IEC 27090 classifies each AI threat by asset, objective, AI type and intent, the four parts of every entry, and the twelve threats.

    Open lesson
  6. Lesson 615 min

    Data and model threats: poisoning, evasion and model theft

    Data poisoning, direct model poisoning, evasion and model exfiltration in ISO/IEC 27090: what each is, its consequences, detection and mitigation.

    Open lesson
  7. Lesson 713 min

    Information exposure threats: inference, inversion and sensitive output

    Membership inference, model inversion and sensitive model output in ISO/IEC 27090, plus Annex B leakage threats: detection and mitigation.

    Open lesson
  8. Lesson 814 min

    Prompt injection and output attacks

    Prompt injection, direct and indirect, and output injection attacks in ISO/IEC 27090: agent risks, detection, mitigation and supply chain.

    Open lesson
  9. Lesson 914 min

    Mitigations and how controls interact (Clause 7)

    Clause 7 of ISO/IEC 27090: why AI security controls must be evaluated together, the ten mitigation themes, the revalidation loop, logging and trade-offs.

    Open lesson
  10. Lesson 1015 min

    Implementing ISO/IEC 27090 in your organization

    How to implement ISO/IEC 27090: AI inventory, extended ISO/IEC 27001 risk assessment, mapping to 27002 controls, templates and a six-month roadmap.

    Open lesson
  11. Final step20 questions · about 30 min

    Final exam and certificate

    Twenty questions drawn from all ten lessons. Score 80 percent or more to issue your certificate with a verification code.

    Go to the exam

Who the course is for

Security leads and CISOs who must extend an ISO/IEC 27001 programme to AI systems. CIOs and technology leaders who own the AI inventory and life cycle. AI product owners, internal auditors, procurement staff and risk managers who need a shared vocabulary for AI security threats. No programming is required.

If you are new to information security standards, start with lesson 1 and read in order. If you already run an ISMS, lessons 5 to 10 carry the material that is new.

How it works

  • Read each lesson. Every one opens with learning objectives and closes with key terms, a one-paragraph summary and questions people ask.
  • Pass the five-question quiz with 4 or more to mark the lesson complete. Every answer comes with an explanation.
  • Take the 20-question final exam. At 80 percent or higher, enter your name and download your certificate as PNG or SVG, with a QR code that decodes its details.

Frequently asked questions

What is ISO/IEC 27090?
ISO/IEC 27090:2026 is an international standard that addresses security threats and compromises specific to artificial intelligence systems. It describes the consequences of those threats across the AI life cycle and explains how to detect and mitigate them. It was published on 14 October 2026 by ISO/IEC JTC 1/SC 27 and runs to 56 pages.
Is this course free, and do I need an account?
Yes. The ten lessons, the quizzes, the final examination and the certificate are free. No account is needed. Progress is stored in your browser, so use the same browser and device to continue where you stopped.
Can I get certified against ISO/IEC 27090?
No. ISO/IEC 27090 is a guidance standard. It contains recommendations rather than auditable requirements, so there is no certification scheme for it. Organizations certify against ISO/IEC 27001 and may add ISO/IEC 42001 for AI management; ISO/IEC 27090 informs the controls inside those systems.
What does the certificate from this course show?
The certificate records that you completed the ten lessons and passed a 20-question examination at 80 percent or higher. It is issued by the course author, Shahzad Asghar, as a record of self-directed learning. It is not an accredited qualification and it is not issued by or affiliated with ISO or IEC.
Who is this course for?
Security leads, CIOs, AI product owners, auditors, procurement staff and anyone who has to decide how an organization protects the AI systems it builds or buys. No programming is required. Familiarity with ISO/IEC 27001 helps but is not assumed.
Does the course replace reading the standard?
No. The course summarises the standard in plain language for learners. The standard itself is copyrighted and sold by ISO and national standards bodies. Teams implementing it should buy the text so they can verify clause wording, Annex A and Annex B.

Related reading on this site

Shahzad Asghar

Senior IT Strategy and Portfolio Officer at UNRWA, with more than twenty years building and securing data and AI systems across UN agencies. He holds the CISM and CISSP security certifications and writes on AI governance, AI security and assurance, and digital transformation. Connect on LinkedIn.

This course is an independent educational resource by Shahzad Asghar. It is not affiliated with, endorsed by or published by ISO or IEC. ISO/IEC 27090:2026 is copyrighted and is sold through the ISO Store and national standards bodies; this course summarises it and does not reproduce its text. Last reviewed: .