GDPR & EU AI Act for SMEs: AI, yes. But done properly.

Most obligations of the EU AI Act have applied since 2 August 2026, and the GDPR (DSGVO) applies anyway. We make both workable in everyday business: clear, audit-proof, without scaremongering.

Request a compliance checkNo legal advice: an honest division of labour with law firms

What applies when, as of November 2026.

No countdown marketing: most of it already applies. Only the deadlines for high-risk AI were postponed in 2026 by an amending regulation.

Four risk classes determine your obligations.

The AI Act sorts AI by the risk of how it is used, not by the technology. The higher the level, the more there is to do. Most applications in a small business sit in the two lower levels.

  1. Unacceptable risk

    Prohibited

    since 2 Feb 2025

    Examples
    Social scoring, manipulative systems, emotion recognition in the workplace and in education, untargeted scraping of facial images.
    What it means
    May not be used at all.
  2. High risk

    High risk

    from 2 Dec 2027

    Examples
    AI in recruitment and personnel decisions, creditworthiness, education and exams, critical infrastructure.
    What it means
    Strict obligations for providers and deployers: risk management, documentation, human oversight, logs.
  3. Transparency risk

    Label it

    since 2 Aug 2026

    Examples
    Chatbots, deepfakes, AI-generated texts meant to inform the public.
    What it means
    People must be able to tell that they are talking to AI or seeing AI content.
  4. Minimal risk

    Free

    no deadline

    Examples
    Spam filters, AI in video games, everyday recommendation logic.
    What it means
    No specific obligations under the AI Act. The GDPR and AI literacy (Art. 4) still apply.

Classification is the first step of the work: it determines what you really need to do. Binding legal information comes from law firms; we provide the technical classification and implementation.

Two workstreams. One system.

GDPR: the foundation

Has applied to practically every business since 2018. What is usually missing is not the will, but the order.

  • RoPAA record of processing activities that is accurate
  • DPAData processing agreements with all service providers
  • TOMsTechnical and organisational measures that are actually followed
  • DPIAData protection impact assessment where needed
  • Art. 15–17Access and erasure with a process instead of panic
  • ConsentConsent and tracking on the website done properly, instead of warning-letter roulette

EU AI Act: the new layer

Builds on the GDPR: if you use AI, you need classification, transparency and trained people.

  • InventoryWhich AI is actually running in your business, including shadow AI?
  • ClassificationEvery application by AI Act risk class
  • Art. 4Measures for AI literacy in the team, training documented
  • Art. 50Labelling AI interaction and AI-generated content
  • Art. 26For high-risk AI: oversight, logs, reporting channels
  • DocsDocumentation that stands up to an audit, not just reassures

Audit-proof in three steps.

  1. Review

    We take stock of data flows and AI use, including the tools nobody officially knows about. No blame, plain language.

  2. Classification & measures

    GDPR gaps and AI Act risk classes are named, prioritised and given honest ballpark figures: what comes first, what can wait.

  3. Implementation, documentation & training

    We implement the technical side, document it audit-proof and train your team (Art. 4 literacy included). Afterwards everything belongs to you. No black box.

A short data protection training in an office: three employees seen from behind at a table, on the whiteboard a process of three boxes with a tick

Implementation

Data protection and AI compliance: in systems, workflows and the team.

Audit-proof, and still easy to follow.

We implement data protection and AI compliance where they arise: in your systems, workflows and team. Not a folder for the shelf, but processes that pass an audit.

Coming soon to the Academy“GDPR & EU AI Act for SMEs”: the course for teams who want to understand it themselves. Join the waiting list with no obligation.To the Academy waiting list

What it costs

  1. Getting started

    the tech check in the initial consultation, 30 minutes, free.

  2. Stocktake and measures

    After that you receive a fixed quote for inventory, classification and measures.

  3. Implementation

    based on the list of measures and the quote.

Your contact

Timur Mert carries out the stocktake and implements the measures. He sets up AI workflows himself, so he knows where data actually flows in practice.

The questions everyone is asking now.

Does the EU AI Act apply to us as a small business?

Yes, as soon as you use AI, even purely as a user (a “deployer”). The heaviest obligations fall on providers of high-risk AI, but transparency obligations and the AI literacy obligation (Art. 4) also apply in small businesses. Since 2 August 2026 this is no longer a transition phase.

Have the AI Act deadlines been postponed?

Partly. Under amending Regulation (EU) 2026/1744, in force since 27 July 2026, the obligations for high-risk AI under Annex III apply from 2 December 2027 and for high-risk AI in regulated products from 2 August 2028. The existing prohibitions and the transparency obligations continue to apply as planned; two new prohibitions (sexualised deepfakes without consent and child sexual abuse material) are added on 2 December 2026. For labelling by generative systems already placed on the market before 2 August 2026, there is a deadline of 2 December 2026. AI literacy (Art. 4) has been reworded: providers and deployers must take measures that support the AI literacy of their staff; they do not have to guarantee a specific level for each individual.

What are the penalties for violations?

Under the AI Act, up to €35 million or 7% of worldwide annual turnover for prohibited practices, up to €15 million or 3% for breaches of other obligations, for example as a deployer or on transparency, and up to €7.5 million or 1% for incorrect or incomplete information supplied to authorities. For companies the higher of the two figures applies, for SMEs and start-ups the lower. Under the GDPR, depending on the breach, up to €10 million or 2%, or up to €20 million or 4% of worldwide annual turnover, whichever is higher. More realistic for small businesses: formal warning letters from competitors, investigations and customers who ask for evidence.

Is it enough to ban ChatGPT in the team?

No. That creates shadow AI: people use it anyway, just without rules. Better: approved tools, clear guard rails, a trained team. That is exactly what we build.

Are you a law firm?

No, and we say so clearly: we implement data protection and AI compliance at a technical and organisational level. For genuine legal questions we work with specialised law firms. No legal advice from us.

Where do we start?

With the review in the tech check: 30 minutes, free of charge. Afterwards you know where you stand, even if the answer is “actually pretty good”.

Sort it now, not explain it later.

Initial consultation free of charge · Review without blame

Software and IT services are provided by Fortes Fortuna LTD OOD (Bulgaria). Details in our legal notice (German)

Initial consultation

Where's the knot?

30 minutes, free. Pick a time that suits you.

with Aylin

The calendar isn't loading right now.

Open in a new tab (opens in a new tab)

Or give us a call: +49 30 83799968

The calendar appears in your browser’s language. Its booking questions are in German.

Rather talk on the phone? +49 30 83799968 · hallo@bemertenswert.de

Calendar: cal.bemertenswert.de · Privacy policy (German)