AI Transparency

Where we use AI

• Blog articles: Research, draft and cover image are created with AI. Every article is read, corrected and approved by Marcel Kuhn before publication.
• Audio versions: The read-aloud function below our articles is a synthetic voice, not a recording of a human being.
• Images: We create cover images and graphics mostly with AI. They show no real people or places unless explicitly stated otherwise. Some scene images on matd.ch that show Marcel Kuhn are created with AI based on a real photo of him. The portraits of him are real photographs.
• Videos on social media: For short videos we use an AI avatar modelled on Marcel Kuhn, together with an AI replica of his voice. The statements are his, the recording is artificially generated.
• Our products: Language models work in the background of Nordstern.Coach, sprichmal.ch and PainToPost.

What a human always does

No article and no video goes online without approval by Marcel Kuhn. Editorial responsibility for everything published here lies with matd ag, not with a model.

How you recognise AI content

We label content where you first encounter it: as a note below the title of an article, as an image caption, right at the play button of the audio version and in the text of every social media post. On every channel the line sits at the end of the post text, before the hashtags; where the platform offers its own AI label (TikTok, YouTube, Instagram, Facebook Reels), we set it as well.

Which tools we use

Language models from Anthropic (Claude), OpenAI and Google (Gemini). Images with Google Gemini, videos with Google Gemini Omni. Speech synthesis with ElevenLabs. Avatar videos with HeyGen. Internally we bundle these tools in our own assistants Ava, Mia and Sam — these are programs, not people.

Questions or corrections

If you come across content that is labelled incorrectly or insufficiently: marcel.kuhn@matd.ch. We will correct it promptly.