AI Act: practical guide for SMEs, what you need to know before August 2026
The European AI Act comes into force on 2 August 2026. Here's what every SME needs to understand and anticipate now. The AI Act in 5 minutes: The European AI regulation is the world's first legislative framework dedicated to AI. It classifies AI systems into 4 risk levels and imposes proportionate obligations. For SMEs, the main challenge is identifying which AI tools they already use, often unknowingly, and which category they fall into. Who is affected?: Any company that uses, develops or deploys an AI system in the EU is affected. This includes SaaS tools embedding AI (CRM with scoring, chatbots, spam filters, writing tools). Even if you didn't develop the AI, you are responsible for its use in your organisation. Concrete obligations for SMEs: For most SMEs, obligations focus on transparency (informing users they're interacting with AI), documentation (maintaining an AI usage register) and governance (designating a responsible person, defining usage rules). How to prepare now: 1) Inventory all your tools embedding AI. 2) Classify them by risk level. 3) Document uses and responsibilities. 4) Train your teams on best practices. 5) Set up a validation process for any new AI tool. The Powehi approach: We support SMEs in their compliance with a framework proportionate to their size: no over-engineering, no panic, but concrete and documented actions.
Key takeaways
- The AI Act applies to any company using AI in the EU
- 4 risk levels: unacceptable, high, limited, minimal
- SMEs must document and govern their AI usage
- A maturity diagnostic is the best starting point
- Deadline: 2 August 2026