AI transparency

1. Where we use AI

• Materials: the exercise texts, tasks, grammar explanations, pictures and audio recordings (synthetic voices) are prepared in advance with the help of AI. They are produced editorially, with no user data of any kind, and we review them before publication — even so they may contain errors.
• AI assessment: your written answers, picture descriptions and short voice recordings are assessed automatically by an AI system. The assessment is a learning aid: it may contain errors, it is not an official exam result, and we give it to nobody but you.
• What AI does not do: it does not decide what you are shown, it does not build a profile of you, and it does not predict how you will do in the exam. Auto-checked exercises (reading, grammar, listening) are marked against a prepared key, without AI.

2. How we mark AI-produced content (Art. 50 AI Act)

• A notice before every interaction with AI: before you submit an answer you are told, at the exercise itself, that an AI system will assess it (Art. 50(1) AI Act). The notice sits at the control you use, not buried in this document.
• A note beside every assessment: "Automated AI assessment — not an official exam result."
• A note about synthetic voices on the listening exercises.
• Machine-readable marking: AI-produced images carry the standardised value "trainedAlgorithmicMedia" in their metadata (IPTC/XMP) and the audio files carry the same value in their ID3 metadata. The AI assessment output is additionally marked in the page itself with an attribute for AI-generated content. Pages carrying such content declare "ai-generated-content" with the same standardised value in the document head, and media files as well as the exercise task files are served with the response header "X-AI-Generated", whose value is that same standardised identifier.
• What is not marked, and why: the brand and identity assets of the Service — the wordmark, the favicon, the home-screen icon and the social preview image — are drawn by the site's own code from a typeface and a plain colour gradient. No model output goes into them and they are not among the directories we mark, so we assert nothing about them either way. The exercise icons, by contrast, were made with AI and are marked like every other image. Nor do we mark this document, the Terms of Use, the Privacy Policy or the Support page: they are not content within the meaning of Art. 50 and the marking does not apply to them.
• Marking text: there is as yet no single technical standard for text. The in-page attribute and the document-head value are our documented best effort under the state of the art, not a claim of a verifiable watermark.

3. Our role and the classification under the AI Act (Annex III, Art. 6)

Our role: we offer the AI assessment system under our own name and are therefore its provider within the meaning of Art. 3(3) AI Act — the fact that it is free of charge changes nothing, because the definition says so expressly. The third-party AI models with which we prepare materials in advance and which assess your answers we merely use; they are supplied by the contracted processor named in the Privacy Policy (OpenAI Ireland Ltd.). We do not name the particular model — it changes, and naming it would publish a detail that goes stale.

The AI Act classifies certain AI systems in education as high-risk (Annex III, point 3(b): systems for evaluating learning outcomes). We checked whether our assessment feature falls under it.

The result of our documented assessment: the feature is not a high-risk system within the meaning of Art. 6(2) AI Act, because the exception in Art. 6(3) applies. It is a voluntary, free aid for self-study; the assessment is purely formative, has no legal or comparable effects, leads to no certificate, decides nothing about access to education or to the exam, and is given to no exam organiser or other party. Only a single, voluntarily submitted answer is assessed; no profiling takes place. The decisive exam is conducted solely by independent human examiners at official exam organisers.

We recorded this assessment in writing under Art. 6(4) AI Act and will present it to the competent authorities on request. It is our own documented assessment, not a confirmation by a lawyer or a supervisory authority. Should the assessment ever change, we would also fulfil the registration duty under Art. 49(2) AI Act.

Since when we mark: the transparency obligations in Art. 50 AI Act (Chapter IV) have applied since 2 August 2026. The amending Regulation (EU) 2026/1744 defers only Sections 1 to 3 of Chapter III (high-risk systems) — the deferral does not reach Art. 50. We therefore mark from the moment the Service is publicly reachable; making it publicly reachable is the placing on the market. We rely on no later date.

4. What we need from you, and what we do not

Please do not put special categories of personal data (for example health, origin, religious belief or political opinions) or other people's personal data into your answers or recordings — we do not need them for the assessment. This is a request, not a filter: the system does not look for such data or separate it out, your whole answer is sent to the processor and it is assessed as a whole. The request we send carries no name, e-mail address or any other user identifier.

Whether and how you learn with our materials is entirely your decision. The AI assessment replaces neither a teacher nor an official exam result — a certificate is obtained only at an exam held by an official organiser. We are an independent private project and are not connected with any examination institution or authority.

Have a question or found a mistake? Write to us at kontakt@slovaka2.sk

5. Read more

• How our materials are made and what we answer for: Terms of Use
• Data processing in the AI assessment, and the operator details: Privacy Policy
• Contact and support: Support