ISO/IEC 42006:2025 – New AI Audit Certification Standard
- Katarzyna Celińska
- Jul 21
- 2 min read
The ISO/IEC 42006:2025 standard has officially arrived — and it’s a step forward for auditing and certifying Artificial Intelligence Management Systems (AIMS). Released on July 7, 2025, this standard sets rules for third-party certification bodies assessing AI systems under ISO42001.
Â
This new framework ensures that audits are credible, consistent, and AI-specific, with enhanced expectations around hauditor competence, transparency, and process structure. It closes a gap between traditional management system auditing and AI governance, including ethics, bias, explainability, and data quality.
Â

Key elements of hasztag#ISO42006:
✅ Based on ISO/IEC 17021-1, it adds AI-specific layers tailored to hasztag#AIMS
✅ Defines audit time estimation formulas and includes templates for certification documents
✅ Covers auditor knowledge requirements, multi-site audits, appeals, surveillance audits, and certification renewal
✅ Emphasizes how certification bodies must ensure impartiality, transparency, and independence
✅ Addresses risk based auditing, AI lifecycle considerations, and the use of external experts in audit teams
Â
This new standard is a valuable tool not just for certification bodies, but also for internal and external auditors looking to ensure effective AI governance. I was amused to see the inclusion of multi-site audit requirements — while they make sense in manufacturing or physical infrastructure, in AI audits, the ‘sites’ are usually data models, algorithms, and documentation. I assume auditors won’t need to ‘see’ AI with their own eyes. Remote tools and video interviews are more than enough!
Â
AI technologies are evolving rapidly — making trust more critical than ever. Assurance through globally recognized frameworks like ISO/IEC 42001 certification, SOC 2 + AI-specific frameworks (e.g., the AI Act), or ISAE3000 attestation offers a highly reliable way to build and demonstrate trust across the entire technology ecosystem.
Â
Author: Sebastian Burgemejster