AI Approval Check: can this AI use case run like this?

The model does not decide, data, purpose and safeguards do. The AI Approval Check assesses a specific use case under the Swiss FADP, the GDPR and the EU AI Act and gives you, in about five minutes, the approval stage, the matching tool architecture and the measures still missing - vendor-neutral, for ChatGPT as much as for Copilot, Claude or Gemini.

How the check works

19 questions in six modules: context, data, purpose and AI Act, product and licence, cross-border transfer, safeguards. The most sensitive data category sets the required protection level, the stage follows from the inherent risk and the gaps in your safeguards.

The result is an approval stage from green (low review effort) through yellow (elevated) and orange (high) to red (stop - not like this), plus protection level 1 to 4 with the matching tool architecture and a prioritised list of measures with their legal basis.

Five questions instead of a vendor question

Typical scenarios from the use case library

What you get

Note

The AI Approval Check is guidance without legal effect. It gives a first assessment based on your answers, is not legal advice, does not replace a legal review of your case and gives no assurance that a use case is permissible. Provider terms and the legal situation change constantly.