Regelgeving

    The EU AI Act for SMEs: What Should You Do if You Use AI? (2026)

    Tergem team

    Published

    Updated

    3 min read

    If you use an AI chatbot, AI receptionist, or AI Operating System in your business, the most important AI Act rule for you is: tell people they are communicating with AI. This transparency obligation has been in place since 2 August 2026. The heavy obligations for high-risk AI have been postponed to 2 December 2027 following the Digital Omnibus.

    This article is for general information and does not constitute legal advice.

    What is the AI Act?

    The AI Act is the European law for artificial intelligence. It came into force on 1 August 2024 and is being implemented in phases. The law categorises AI based on risk: the greater the potential harm to people, the stricter the rules.

    The four risk levels

    LevelExamplesWhat applies?
    ProhibitedSocial scoring, manipulative AI, emotion recognition in the workplaceNot allowed
    High riskAI in recruitment and selection, credit scoring, educationStrict requirements, from 2 December 2027
    TransparencyChatbots, AI receptionists, AI-generated contentMust state it is AI
    MinimalSpam filters, text suggestions, internal analysesNo specific requirements

    Most SME applications fall into the last two categories.

    The timeline after the Digital Omnibus

    DateWhat comes into effect
    2 February 2025Prohibited AI practices
    2 August 2025Rules for providers of general-purpose AI models
    2 August 2026Transparency obligation (declaring communication with AI)
    2 December 2026Watermarking for AI content from systems already on the market
    2 December 2027High-risk AI from Annex III, such as recruitment and selection
    2 August 2028High-risk AI in regulated products (Annex I)

    The Digital Omnibus on AI came into force on 27 July 2026. Many online articles still mention August 2026 for high-risk; this is no longer correct.

    What should you do as an SME now?

    1. Create a list of your AI applications

    Note which AI tools you use, for what purpose, and with what data. Also consider AI in software you already have, such as your CRM or accounting package.

    2. Declare AI upfront

    Do customers talk to a chatbot or AI receptionist? Have it state at the beginning that it is an AI assistant, for example: "You are speaking with the digital assistant of [company]." Also include this in your privacy policy.

    3. Check if you are doing anything high-risk

    Do you use AI to select job applicants, evaluate employees, or determine creditworthiness? Then prepare for the requirements that apply from December 2027: documentation, human oversight, and risk management.

    4. Train your team

    Ensure employees know what AI can and cannot do, what data they should not input, and when a human needs to step in. The AI Act pays attention to AI literacy; ask your advisor what specific obligation applies to you following the Omnibus.

    5. Combine with the GDPR

    The AI Act is in addition to the GDPR, not a replacement for it. If you process personal data with AI, the GDPR rules still apply. Read more about this in AI and privacy: using customer data safely.

    How high are the fines?

    For prohibited AI, fines can reach up to €35 million or 7% of global annual turnover; for other violations, up to €15 million or 3%. For SMEs, the lower of these two amounts always applies. Supervision in the Netherlands is carried out by, among others, the Dutch Data Protection Authority (Autoriteit Persoonsgegevens) and the Dutch Authority for Digital Infrastructure (Rijksinspectie Digitale Infrastructuur).

    How Tergem handles this

    Our AI receptionist states at the beginning of every conversation that the caller is speaking with an AI assistant, data remains in the EU where possible, and every customer receives a data processing agreement. Want to know more? View our services or ask your question.

    Frequently asked questions

    Yes. The AI Act applies to everyone who develops or uses AI in the EU. For SMEs, however, lower maximum fines and supporting measures apply.

    Yes. Since 2 August 2026, people must be informed that they are communicating with an AI system, unless this is already obvious to everyone.

    No. Writing or summarising texts, or using a chatbot for customer queries, does not fall into the high-risk category. High-risk refers to applications such as recruitment and selection, education, credit scoring, and critical infrastructure.

    An EU amendment to the AI Act that came into force on 27 July 2026. It shifts, among other things, the high-risk obligations to December 2027 and August 2028.

    Related articles

    Book a free demo

    Discover what an AIOS or AI receptionist can do for your business. Free, with no obligation.

    Book a free demo