T-Matrix
EN
Get a quote

Legal

AI Transparency

We use AI in almost everything we do. Here is where, who checks it, and how you can tell AI-generated content apart.

DraftThis text is a draft and is being reviewed before launch. It describes our internal AI policy, which is not yet formally adopted.

This page follows Article 50 of the EU AI Act (Regulation (EU) 2024/1689), applicable since 2 August 2026, and our internal AI policy.

In short

  • A human checks AI output before it reaches a client or the public.
  • AI does not make final decisions about people.
  • Chatbots we build say that they are AI. Deepfakes we produce carry a visible EU label.
  • We currently operate no high-risk AI systems.

AI in our work

We use AI for software development, translation, marketing texts and AI production. Rules we follow:

  • Only approved AI services, through business accounts or APIs that do not train on our inputs.
  • Client data and personal data go only to services with a data processing agreement, preferably hosted in the EU, or after pseudonymisation.
  • Passwords, keys and other secrets never go into AI requests.
  • Code goes through human review before release. Texts, translations and media are checked by the responsible person.
  • AI agents work with their own keys and limits. Deployments, payments, deletions and outgoing messages need a human approval.

On this website

  • Texts are written with the help of AI and edited by our team, who take editorial responsibility for them.
  • The animated dots in the background are drawn by our own code in your browser. They are not AI-generated media.
  • AI-generated images, video or audio on this website will carry the EU label for AI-generated content.
  • Chat assistant: there is none yet. If we add one, it will tell you at the first message that you are talking to an AI.
  • Translations into other EU languages will be made with AI and reviewed by native speakers for key pages. For legal pages the German text is binding.

AI video, audio and music

In our Create services we generate content with AI tools for clients. Here we are the deployer and we label the content:

  • Every deepfake (for example an AI avatar, a cloned voice or realistic AI footage) carries a visible label from the first moment it is shown, embedded in the content itself so it survives download and resharing.
  • We use the EU icons Fully AI-Generated, Partially AI-Modified and Basic. Artistic or satirical works get a label that does not disturb the work.
  • Audio without picture starts with a spoken note, for example "This audio was created with AI."
  • We use a real person's face or voice only with their written consent, collected before production.
  • We keep the machine-readable marks of AI tools and log every labelled order. A second person checks each label.
  • Clients may not remove or cover the labels (see Acceptable use).

Our products

  • HyperSale is our own product, so we are its provider under the AI Act. It scores companies, not people; a human confirms every score. It drafts letters and personal pages with AI. Users review the first letters and can then switch on automatic sending. Machine-readable marking of AI-written texts (Article 50(2)): [to be added: current status]
  • Stellwerk records and limits what AI agents do: masking of personal data before cloud AI, approvals for risky actions, a journal for AI Act records. Its Guard module monitors agents, never the behaviour or emotions of employees. Its AI Act classification will be published before market launch. [to be added]

High-risk check

We keep a register of all AI systems we use, build and operate, with their risk class. Our current result: T-Matrix operates no high-risk AI systems. For client projects, every brief asks whether the AI will make or support decisions about people; if yes, we run a high-risk check before we make an offer.

Team training

Everyone who works with AI at T-Matrix completes our internal AI literacy course within the first month and a refresher every year, as Article 4 of the AI Act asks. We record who completed it and when.

Voluntary commitments

We intend to join the EU AI Pact and to sign the EU Code of Practice on transparency of AI-generated content (section 2, for deployers). Both are voluntary. We will state it here only once we appear on the official lists of the European Commission, with a link so that you can check it yourself. [status: not yet signed]

Questions

Questions about our use of AI or a label you found? Write to hello@t-matrix.net. Our compliance work is technical and organisational support, not legal advice.