Playbook · for Leadership, Security
EU AI Act and ISO 42001: a starter for small teams
Find out which AI Act role you have, which dates apply, and how to build the evidence once with ISO 42001 and ISO 27001.
By Mr.DLast reviewed 3 min read
Start with your role
The EU AI Act (Regulation (EU) 2024/1689) gives different duties to different roles. Most small companies are deployers: they use AI systems built by others. You are a provider if you develop an AI system, or have one developed, and put it on the EU market under your name, which includes AI features inside your SaaS product. Importers and distributors have their own, lighter duties. Many companies are both deployer and provider.
Then find your risk level
| Level | Examples | What it means |
|---|---|---|
| Prohibited | social scoring, manipulative techniques, some biometric uses | Banned since 2 February 2025 |
| High-risk | AI in hiring, credit scoring, education, critical infrastructure, or safety components of regulated products | Risk management, data governance, logging, human oversight, accuracy, cybersecurity, documentation, conformity assessment |
| Transparency | chatbots, AI-generated or manipulated content | Tell people they deal with AI; mark synthetic content |
| Minimal | spam filters, most internal productivity uses | No new duties beyond AI literacy |
Providers of general-purpose AI models have had their own obligations since 2 August 2025.
The dates after the Digital Omnibus
The Digital Omnibus on AI (Regulation (EU) 2026/1744, in force since 27 July 2026) moved the high-risk dates:
- 2 December 2026: marking of AI-generated content applies to systems placed on the market before 2 August 2026 (other transparency duties already apply).
- 2 December 2027: obligations for stand-alone high-risk systems listed in Annex III.
- 2 August 2028: obligations for high-risk AI built into products covered by EU product law (Annex I).
The delay changes when the duties apply, not what they are. Building the evidence takes longer than the remaining time suggests.
Five steps for this quarter
- Inventory every AI system you build or use, with an owner, purpose, data used and vendor.
- Classify each by role and risk level, and record why. Most will be minimal risk; write that down too.
- AI literacy: take measures so staff and others who operate AI for you understand its limits; since the Omnibus this is a duty of effort, so keep training records as proof of the effort.
- Transparency: label chatbots and mark AI-generated content where the Act requires it.
- Gap-check high-risk candidates early, because their documentation, logging and human oversight take months.
Where ISO 42001 fits
ISO/IEC 42001:2023 is the certifiable management system for AI. It gives you the structure the Act assumes you have: an AI policy, roles, risk and impact assessments, lifecycle controls, supplier oversight, monitoring and improvement. If you already run an ISO 27001 ISMS, you reuse its clauses for context, leadership, documents, internal audit and management review, and add the AI-specific parts. ISO 42001 certification does not by itself prove AI Act conformity, but it produces much of the evidence.
And the NIST AI RMF
The NIST AI Risk Management Framework (Govern, Map, Measure, Manage) and its Generative AI Profile are free, practical and widely referenced by US customers. Use them as the working method for the risk assessments that ISO 42001 and the AI Act ask for.
How WardBee helps
WardBee already automates the technical evidence the Act and ISO 42001 share with ISO 27001: access control, logging, encryption and secure development across AWS, GitHub and GitLab, plus policies and manual controls for the rest. ISO 42001, EU AI Act and NIST AI RMF are guidance in WardBee today; automated control mapping for them is coming. This page is not legal advice; check the regulation text for your case.
Related standards: EU AI Act · ISO/IEC 42001 · NIST AI RMF


