AI transparency & EU AI Act

Honesty is at the heart of StoffwechselFit – and that includes saying openly where artificial intelligence (AI) helps us and where it does not. This page is our transparency statement under Article 50 of the EU AI Act (Regulation (EU) 2024/1689), written in plain language and without jargon.

Principle

For us, AI is a tool, not an author. It helps with drafting, organising, translating and illustrating. Responsibility for every published statement, image and figure rests with people. Nothing goes online that a human has not read, checked and approved first. You therefore never read “raw” AI output but editorially controlled content – and the labelling shows you where AI was involved.

Where AI helps

• Illustrations: The flat, calm vector illustrations for categories, articles and tools are created with AI assistance according to our style guidelines. They show still lifes, curves and symbols – no close-ups of people, no brands, no measured values.
• Recipe images: Photorealistic recipe images are generated with AI assistance and then checked by us for plausibility: do ingredients, portion and presentation match the recipe? Only then is the image approved.
• Translation drafts: The English version of an article often starts as an AI draft. A person compares it paragraph by paragraph with the German original, corrects technical terms and units, and approves it.
• Structure suggestions: When outlining articles, wording subheadings, drafting tables and writing alternative text for images, we use AI as a sparring partner. The editorial team decides what stays in the text.

Where AI does not decide

• Approval by people: Whether an article, a recipe or an image is published is decided exclusively by a person via the approval page of our editorial system. There is no automatic publishing.
• Medical statements: Every health-related statement is checked against the cited sources and given an study context. AI output is never treated as research. There is no AI chatbot answering your health questions, and no automated recommendations on diet, medication or treatment.
• Measurement data: The CGM values in the Field Lab articles come from the author’s real sensor data. They are neither generated nor altered by AI. The calculators on this website use disclosed formulas, not AI. What you enter into trackers and calculators stays in your browser and is not analysed.

How we label

So that you never have to search for AI involvement, we label it on four levels:

• Visible “KI-Bild” badge: AI-assisted article header images and recipe images carry a small, clearly legible “KI-Bild” (AI image) badge directly on the image. Small icon graphics, for example on the tool tiles, are marked as AI-assisted in the page markup (data-swf-ai-assisted) but carry no separate badge.
• Meta field swf_image_source: The media library records for every image how it was created, for example “AI-assisted; editorially reviewed” or “own photograph”.
• HTML attribute data-swf-ai-assisted="true": In the page code, AI-assisted images are marked in machine-readable form so that browser extensions, search engines and accessibility tools can read the label as well.
• Notice in the page footer: Every page links in its footer to this transparency statement and states that texts are produced with AI assistance and editorially reviewed.

Our own photographs, screenshots of sensor data and charts drawn from measured values carry no AI badge; they are identified separately as original material.

Review process

Every piece of content goes through the same three steps before publication:

• Expert approval: A person checks sources, figures, units, study contexts and safety notes and sets the status “reviewed – approved”. Without this status the article does not appear.
• Content hash: At the moment of approval a checksum (hash) is calculated over text and images. If the content changes afterwards, the system detects the deviation and the article must be approved again.
• Revision log: Approvals, changes and corrections are logged with a date. Articles carry a “last reviewed” date, sources a “sources as of” date. We correct errors visibly, not silently.

How we read and assess measurement data is described in detail on our page Methodology & review process.

Risk assessment

StoffwechselFit is an information and documentation website. In our assessment, the use of AI here involves low risk: we operate no AI systems that make decisions about people, no automated diagnosis, no profiling based on health data, no emotion recognition and no biometric analysis. AI-generated images are labelled, and AI-assisted texts are editorially accounted for. We repeat this assessment as soon as an interactive AI feature, an automated analysis of personal data or an automated recommendation were to be added. We deliberately do not display a blanket “AI Act compliant” seal; instead we describe what we actually do.

Responsible entity

Responsible for this statement and for the use of AI on stoffwechselfit.com is the operator of the website, SSN Computer GmbH & Co. KG, Peter-Fix-Straße 37, 53474 Bad Neuenahr-Ahrweiler, Germany. Articles are published under the pseudonym StoffwechselFit; the person responsible for editorial content is named in the legal notice.

Contact

Would you like to know how a particular piece of content was created, or have you spotted an error? Write to us via the contact page. We are happy to explain which steps an article went through, and we correct whatever needs correcting.

Version: 2 Nov 2026