The European Commission says the AI Act transparency obligations apply from 2 August 2026. For banks and insurers, the practical task is to identify where customers, employees or the public interact with AI-generated or AI-assisted content and then make the required disclosure clear, timely and understandable.
Inventory the touchpoints
Review chat assistants, claims communications, marketing material, synthetic images, voice systems, document drafting and public-interest content. A tool used only internally may still create disclosure or record-keeping issues if its output is sent externally without meaningful review.
Design disclosure into the workflow
Disclosure should be visible at the point of interaction, not buried in a general privacy policy. Record the wording, trigger, language, responsible owner and evidence that the control operated. For generated documents, retain the source and reviewer so a recipient can be told accurately how the content was produced.
Test the exception cases
Test translation, accessibility, customer complaints, human-edited output and mixed human-AI content. The Commission's transparency code and AI regulatory framework should be read with the exact system role and use case.