Regulation (EU) 2024/1689 · In force

The AI Act applies to your SME.
Here is what you need to know — and do.

The European AI Act is the world's first regulatory framework dedicated to artificial intelligence. It concerns any organization using an AI tool in Europe — including third-party SaaS software. Themio identifies your obligations and structures your compliance, without an in-house lawyer.

The essentials in 4 points

The 4 key points of the AI Act for your business

Risk-based approach

4 classification levels: Unacceptable, High, Limited, Minimal. Your level determines your exact obligations.

Every SME is affected

If you use a third-party AI tool, you are a "deployer" and have compliance obligations.

New deadline: December 2, 2027

The compliance deadline for high-risk systems is extended to December 2, 2027. Note: compliance remains mandatory, only the timeline has changed.

Fines up to 7% of global turnover

In case of violation of prohibitions (unacceptable risk systems), penalties are immediate and severe.
What is the AI Act?

AI Act, AIA, Regulation (EU) 2024/1689: one text, multiple names

The AI Act — also called AIA (Artificial Intelligence Act) or Regulation (EU) 2024/1689 — is the European legislation that governs the development and use of artificial intelligence systems within the European Union. Published in August 2024, it is the world's first comprehensive regulatory framework dedicated to AI.

Its goal is twofold: to ensure trustworthy AI that respects fundamental rights and privacy — while preserving space for innovation. By legislating first, Europe is taking a global strategic lead on this issue.

Key Takeaways

Who is concerned?

Your SME is concerned — even if you didn't develop the AI yourself

The regulation applies to any organisation that develops OR uses an artificial intelligence system in Europe. Under the official definition, an AI system is any automated system designed to generate predictions, recommendations, decisions, or content that influence real or virtual environments.

In practice: a SaaS software with an AI feature falls under the regulation as soon as it influences a decision or automates a workflow — even if you didn't write a single line of code.

Provider or deployer: what's the difference?

Provider Deployer
Develops or places an AI system on the market Uses an AI system developed by a third party
E.g.: vendor of an automated HR scoring tool E.g.: SME using that tool to recruit
Maximum obligations — documentation, compliance, registration Verification, monitoring, and transparency obligations
Few SMEs in this category The vast majority of SMEs are deployers

Important note: If your SME uses a CRM with AI scoring, an automated recruitment tool, or a decisional chatbot, you are a deployer — and you have obligations. To audit and map these undeclared AI usages across your teams, check our deep-dive on Shadow AI in SMEs .

The Risk-Based Approach

The 4 AI Act Classification Levels

Level 1: Unacceptable risk
PROHIBITED — IN FORCE SINCE FEB. 2025

Prohibited systems: immediate sanctions

AI systems presenting unacceptable risks have been strictly banned since 2 February 2025. They are incompatible with European fundamental rights.

Specifically banned:

  • Subliminal manipulation systems that distort human behavior
  • Social scoring (behavioral grading of individuals)
  • Real-time biometric identification in publicly accessible spaces (with narrow public safety exceptions)
  • Systems exploiting vulnerabilities of specific groups
Alert: If your organisation uses any of these systems, you face immediate sanctions. There is no transition period — the ban is fully in effect.
Level 2: High risk
STRICT OBLIGATIONS — NEW DEADLINE: DECEMBER 2, 2027

High-risk systems: mandatory compliance by December 2, 2027

Digital AI Omnibus in effect: following the entry into force of the Digital AI Omnibus (27 July 2026), the compliance deadline for high-risk systems (Annex III) is confirmed as 2 December 2027. The timeline is extended — the obligations remain.

This is the most impactful level for SMEs and mid-caps. It covers use cases already deployed in many business functions. The regulation identifies 8 high-risk areas ( Annex III of the Regulation ):

Domain Concrete example for an SME
Recruitment & HR Automated resume screening, candidate scoring
Credit & Insurance Financial scoring or creditworthiness evaluation
Healthcare Automated medical diagnostic support
Education Automated student grading or evaluation
Critical Infrastructure Automated energy or transport network management
Justice Judicial decision support tools
Law Enforcement Automated surveillance systems
Migration Automated border control systems

What obligations apply to your company before December 2027?

  • Comprehensive technical documentation of the AI system
  • Conformity assessment before deployment
  • Registration in the EU high-risk AI system database
  • Effective and auditable human oversight
  • Fundamental Rights Impact Assessment (FRIA) — applicable to public bodies and essential service operators

Good news for SMEs: the Omnibus extends SME simplifications to small mid-caps (< 750 employees, < €150M turnover), with streamlined documentation templates and priority access to regulatory sandboxes.

SME Alert: If you use an automated resume screening tool, credit scoring, or any automated evaluation system in HR or finance, you are affected. Themio determines your risk level in a few minutes.
Level 3: Limited risk
TRANSPARENCY OBLIGATIONS

Limited risk systems: inform users

These systems do not present major structural risks, but must adhere to transparency requirements. Users must be informed that they are interacting with an AI.

Examples:

  • Chatbots and conversational assistants
  • Content generation tools (text, image)
  • Commercial recommendation systems
Key obligation: Clearly inform users that they are interacting with an AI — particularly on your website, app, or customer service ( Article 50 ). Benchmark your Article 50 compliance in 3 minutes →
Level 4: Minimal risk
NO SPECIFIC OBLIGATIONS

Minimal risk systems: free use

The majority of general-purpose AI applications fall into this category. They are not subject to any specific regulatory obligations under the AI Act, although the GDPR and other regulations may still apply.

Examples:

  • Spam filters
  • Standard automatic translation tools
  • Non-decisional content recommendations
Timeline

AI Act Timeline: key deadlines you cannot miss

Date Obligation Status
August 2024 Publication of the regulation in the Official Journal of the EU
In effect
February 2025 Prohibition of unacceptable risk systems (Level 1)
In effect
August 2025 Obligations for General Purpose AI models (GPAI)
In effect
27 July 2026 Digital AI Omnibus formally adopted — entered into force
In effect
2 August 2026 Article 50 transparency obligations (AI-generated content labeling, chatbot disclosure) — not postponed
In effect
Dec. 2, 2027 Mandatory compliance for high-risk systems Annex III (HR, credit, education...)
Confirmed — postponed
Aug. 2, 2028 Full application — high-risk systems Annex I (medical devices, machinery...)
Confirmed — postponed

The Digital AI Omnibus is now law. Formally adopted by Parliament (16 June 2026) and the Council (29 June 2026), signed 8 July 2026, and in force since 27 July 2026, it confirms the postponed deadlines: 2 December 2027 for standalone high-risk systems (Annex III), and 2 August 2028 for high-risk AI embedded in regulated products (Annex I). Note: Article 50 transparency obligations were not postponed and have applied since 2 August 2026. The obligations have not disappeared — only the calendar changed. This is still the right window to start your compliance before deadline pressure returns. Themio determines your risk level in a few minutes.

What Themio does for you

Themio automates your AI Act compliance — without an internal legal team

The AI Act imposes a structured compliance process: classifying your systems, documenting your obligations, and producing evidence for the regulator. Themio transforms this process into an automated workflow. For a step-by-step roadmap, explore our 5-step compliance guide and our 2026 compliance tools benchmark .

Classification of your AI systems

Themio determines the risk level of each AI system you use or deploy — via a deterministic rules engine, not a guessing chatbot. Result: Unacceptable, High, Limited, or Minimal, with an article-by-article explanation.

→ Auditable, explainable result, citable to a regulator

Mapping of your exact obligations

Once your risk level is established, Themio lists the precise obligations that apply to you — based on your role (provider or deployer), your sector, and your systems. No generic lists. Your obligations, linked to the applicable article.

→ Prioritized action plan, exportable as a PDF

Generation of your governance documents

Risk management plans, AI governance policies, system registers, transparency notices — generated automatically and ready to present to your regulator or investors.

→ Ready for audit from day one
Resources & Practical Guides

Deepen Your AI Act Compliance with Expert Analysis

Methodological guides, regulatory breakdowns, and software benchmarks to navigate European AI compliance step-by-step.

Free Self-Assessment Tool

Article 50 AI Transparency Benchmark

Test your conversational chatbots, assistants, and generated media to verify whether your user disclosures satisfy mandatory Article 50 requirements.

Launch Benchmark (3 min) →
FAQ

Frequently Asked Questions about the AI Act and SME compliance

What is the AI Act and who does it apply to?
The AI Act (EU Regulation 2024/1689) is the world's first comprehensive legislation governing the use of artificial intelligence. It applies to any organization that develops or uses an AI system in Europe — including SMEs that use third-party SaaS software integrating AI features. If your company is based in the EU or if you offer services to EU users, you are affected.
My company uses ChatGPT or a CRM with AI — am I subject to the AI Act?
Yes, potentially. As soon as a tool you use meets the definition of an "AI system" under the regulation (generating predictions, recommendations, or automated decisions), you are considered a "deployer" and have obligations. The level of your obligations depends on the classification of the system involved (High risk, Limited, or Minimal). See how Themio classifies your systems in minutes.
What are the penalties for non-compliance with the AI Act?
Fines vary according to the severity of the infringement. The use of an unacceptable risk system (prohibited since February 2025) is punishable by fines of up to €35 million or 7% of total worldwide annual turnover (whichever is higher). Other violations are penalized up to €15 million or 3% of global turnover. Proportionate sanctions are provided for SMEs, but the compliance obligation remains fully in force.
What is the difference between the AI Act and the GDPR?
The GDPR regulates the processing of personal data. The AI Act regulates artificial intelligence systems. The two often overlap: an AI system that processes personal data is subject to both regulations simultaneously. Themio analyzes your documents against both frameworks in the same audit, avoiding any blind spots.
How can I determine if my AI system is "high risk"?
The regulation defines 8 high-risk areas in its Annex III: recruiting and HR, credit and insurance, health, education, critical infrastructure, justice, law enforcement, migration. If your organization uses an AI system in one of these areas — even via a third-party tool — you likely fall into the high-risk category. Themio determines your classification via a rules engine built article by article, with an explainable and auditable result.
How long does it take to become compliant with the AI Act?
It depends on the number of systems involved and their risk level. The first step — classifying your systems and identifying your obligations — can be completed with Themio in minutes. Full compliance implementation (documentation, internal processes, governance) generally takes from a few weeks to a few months depending on your organization's complexity. With the August 2026 deadline for high-risk systems, the time to act is now.

Identify your AI Act obligations in under 2 minutes.

Themio classifies your AI systems, maps your obligations article by article,
and generates your governance documents — without an internal legal team or costly consultants.

🔒 EU Hosting · Data not shared · No commitment