Enable the user to access an explanation of why the AI system behaved as it did.
Limit disruptive changes when updating and adapting the AI system’s behaviors.
Inform the user when the AI system adds or updates its capabilities.