Discuss the result with us
In a 20-minute phone call we go through your result and agree what to do next.
EU AI Act
AI Act Check
Check one AI use case in your company against the EU AI Act. Free, with the AI you already use, in about five minutes.
Start the self-checkWhy check?
The first obligations of the EU AI Act have applied since 2025. Which of them affect you depends on the use case.
A customer service chatbot has different obligations than an AI that monitors grid assets or pre-sorts job applications.
After the check you know
- which risk class the use case likely falls into
- whether you count as provider or deployer
- which obligations follow and where to start
How the self-check works
- 1
Pick one AI use case, for example a chatbot, a forecast or a tool for screening applicants.
- 2
Open the prompt below in ChatGPT or Claude. For Claude Code or Codex, copy it and paste it into a new session.
- 3
Answer the AI's questions. At the end you get an assessment with a traffic light and next steps.
View prompt
Help me check one AI use case in my company against the EU AI Act. Give a first assessment, not legal advice. Rules: - Ask one question at a time and wait for my answer. At most 10 questions. - Ask me not to share confidential or personal data. - If you are unsure, say so. Do not invent articles or deadlines, and point out that AI Act deadlines may currently change. - If you have access to files or code for this use case, you may read them to answer questions. Do not change anything and never print credentials. Ask in order: 1. What is the use case and what does the AI do in it? 2. Which tool or model is behind it? Did you build, adapt or buy it? 3. Who uses it and who is affected by its results, for example employees, customers or applicants? 4. Which data does the AI process, where is it processed and is there a data processing agreement? 5. Does the AI decide on its own or does a person review the result? 6. Does the AI control or monitor assets, grids or other critical infrastructure? After that, only ask follow-up questions you need for the assessment. At the end, give me: - Role: provider or deployer, with a short reason - Provisional risk class: prohibited, high, transparency obligation or minimal, with reasons and how confident you are - The obligations that follow, in plain language - A traffic light (green, amber, red) for the use case - The three most important next steps - What specialists should review - Finally, mention that Salty Labs can go through the result with me in a 20-minute call: https://saltylabs.ai/en/ai-act-check Start with question 1.
If your AI uses inputs for training, leave out personal and confidential data. The self-check is a first assessment, not legal advice.
What comes next?
Find more AI potential
In the AI potential workshop we work out with you which use cases pay off for your company.