Assessment questionnaire

What is the name of your AI system?

This name will be used to identify your system throughout the compliance report.

Which of these scenarios best describes your system's behaviour?

This question determines whether your system qualifies as AI under the regulation. A rules-based or simple statistical system is not covered by the AI Act.

It applies fixed rules written in advance (e.g. if amount > €10,000, alert)It calculates averages or totals (analytical dashboard)It was trained on historical data and infers results from new data (scoring, classifier)It generates new content (text, images, code, audio)I'm not sure about the technical behaviour (hint: if the system adjusts its outputs from data without being reprogrammed, it's probably AI)

Does your system use a foundation model or a large language model (LLM)?

Examples: GPT, Claude, Mistral, Gemini, Stable Diffusion. The type of model used determines your obligations towards the AI provider and your level of responsibility.

Yes - via an external API (OpenAI, Mistral, Anthropic, Google…)Yes - we fine-tuned or adapted a pre-trained modelYes - we host an open-source model internallyNo - model trained from scratch in-house (classical ML, specialised deep learning)No - this is not machine learning

Is the system being used for a purpose different from the one it was designed for?

If you use an AI tool for purposes other than what it was originally intended for, you may become legally responsible for its compliance as if you had designed it yourself.

No - used exactly as intended by the providerThe use has been extended but remains in the same domainYes - use not intended by the original providerWe designed the system for our own use