- ← Retour aux ressources
- /EU AI Act: The SME Guide for 2026
EU AI Act: The SME Guide for 2026
Timeline, risk levels, obligations: what SMEs need to understand about the European AI regulation before the 2 August 2026 deadline.

One date is making its way into the calendar of every European business: 2 August 2026. That is when the European Union's regulation on artificial intelligence, the EU AI Act, becomes fully applicable. If your SME uses a chatbot, a CV-screening tool, a content generator or any kind of "smart" software, you may be concerned, and it is worth understanding why today.
In brief
- The EU AI Act entered into force on 1 August 2024 and becomes fully applicable on 2 August 2026 (with some exceptions), according to the European Commission.
- Prohibited AI practices and AI-literacy obligations have already applied since 2 February 2025.
- Obligations for providers of general-purpose AI (GPAI) models have applied since 2 August 2025.
- From 2 August 2026, the European Commission's enforcement powers for GPAI providers apply, along with transparency rules (for example, labelling AI-generated content).
- The regulation sets out a four-tier risk-based approach and penalties of up to €35 million or 7% of total worldwide annual turnover for the most serious infringements.
- Most SMEs are users (deployers) of limited- or minimal-risk AI: their obligations are real but proportionate.
General information for educational purposes. This guide is not legal advice. For your specific situation, consult a qualified professional.
What is the EU AI Act?
The EU AI Act is the world's first comprehensive legal framework on artificial intelligence: a European Union regulation that governs how AI systems are placed on the market and used, based on the level of risk they present.
Its core principle is a risk-based approach: the more an AI system could harm people's safety or rights, the stricter the obligations. A spam filter and a credit-scoring system are not treated the same way, and that distinction is exactly what this classification is about.
The 4 risk levels
The regulation sorts AI systems into four categories. Here is a simplified overview.
| Risk level | Examples | Main obligations |
|---|---|---|
| Unacceptable (banned) | Social scoring by public authorities, behavioural manipulation, certain forms of biometric recognition | Practices prohibited in the EU |
| High | AI for recruitment, access to credit, certain medical devices or critical infrastructure | Strict obligations: risk management, data quality, documentation, human oversight |
| Limited | Chatbots, conversational assistants, image or text generators | Transparency duties: inform users they are interacting with an AI, label generated content |
| Minimal | Spam filters, AI in video games, basic suggestions | No specific obligation |
The vast majority of tools SMEs use day to day fall under the limited or minimal levels.
The timeline to know
The EU AI Act is being applied in stages. Here are the milestones to remember, as set out by the European Commission.
1 August 2024
Entry into force
2 February 2025
Prohibited practices
2 August 2025
General-purpose models (GPAI)
2 August 2026
Full application
2 August 2027
GPAI placed on the market before August 2025
Note: a simplification package nicknamed the "AI omnibus" was adopted on 19 November 2025, followed by a political agreement reached on 7 May 2026. Some rules may still be adjusted, so it is wise to follow official communications.
Am I concerned as an SME?
Yes, but probably not in the way you fear. The regulation distinguishes between providers (those who develop and place an AI system on the market) and deployers, that is, professional users.
Most SMEs are deployers of limited- or minimal-risk AI tools. Your main responsibilities are therefore reasonable:
- Transparency: If you use a customer-facing chatbot or publish AI-generated content, you must make that clear.
- AI literacy: You must ensure that the people using these tools have a sufficient level of understanding of how they work and what their limits are.
- Vendor due diligence: Choose providers that themselves comply with the regulation. Their compliance becomes part of yours.
Penalties do exist, up to €35 million or 7% of total worldwide annual turnover for the most serious infringements, such as the use of prohibited practices. But the penalty regime is designed to be proportionate: a small business using a chatbot transparently is not in the same situation as a large provider of high-risk models.
What to do right now
You do not need a legal department to make progress. Here is a practical checklist to get your SME ready.
- Inventory your AI uses. List every "smart" tool actually in use: chatbots, writing assistants, application screening, scoring, image analysis, and so on.
- Classify each use by risk level. For each tool, ask: unacceptable, high, limited or minimal? Most will fall into "limited" or "minimal".
- Check your vendors' compliance. Ask your providers how they comply with the EU AI Act, and keep their answers in writing.
- Train your staff. Put in place simple AI-literacy awareness: what the tool does, what it cannot do, and when to keep a human in the loop.
- Document. Keep a written record of your inventory, your classifications and your decisions. When it comes to compliance, what is not documented does not exist.
FAQ
Does the EU AI Act apply to my small business?
Very likely, as soon as you use AI tools. The good news: as a deployer of limited- or minimal-risk AI, your obligations mainly concern transparency, training and choosing compliant vendors.
What happens if I do nothing by 2 August 2026?
You expose yourself to a non-compliance risk. Penalties can reach €35 million or 7% of worldwide turnover for the most serious cases, but they are designed to be proportionate to the size of the business and the seriousness of the infringement.
Do I have to label AI-generated content?
Transparency rules apply from 2 August 2026 and include, among other things, informing people when they interact with an AI and labelling certain generated content. Adopting these good practices now puts you ahead.
What is the "AI omnibus"?
It is a simplification package adopted on 19 November 2025, with a political agreement reached on 7 May 2026. It aims to clarify and lighten certain rules. As the matter is evolving, follow the European Commission's communications.
Conclusion
The EU AI Act is not a mountain reserved for tech giants. For most SMEs, compliance comes down to a few common-sense steps: knowing which tools you use, understanding their risk level, choosing serious vendors and training your teams. The European Commission remains the reference source for tracking how the regulation evolves.
The 2 August 2026 deadline is the perfect opportunity to bring order to your AI uses, calmly and without rushing. To go further, explore our resources on AI and automation for SMEs.