The examination and validation should include the assumptions, operational protocols, data properties and output decisions of the AI technology.

Ultimately, such work should be validated by regulatory agencies or other supervisory authorities.