Lesson · 6 min · Stage 6
Transparency and explanations
Learning goal: Decide what to tell people about AI, and how.
For reference purposes only. Not legal advice.
Draft: not yet fact-checked or legally reviewed
Key points
- Tell people when they are dealing with AI or AI-generated content.
- Explain in terms people understand: what the system does, its limits, and how to reach a human.
- Give people a route to contest decisions that affect them.
- Some regions require disclosure, labelling or notice; rules differ and some are not settled.
- Transparency inside the organization matters too: model cards, data records and logs.
Who this is for: anyone new to the topic. Related stage: Approve and deploy.
Two kinds of transparency
Outward transparency is what you tell users and affected people. Inward transparency is what you can show your own organization, auditors and regulators: what the system is for, how it was tested, where its data came from, and who is responsible.
What to tell people
- That AI is in use. A person chatting with a bot should know it is a bot. People seeing AI-generated images, audio, video or text may need a label.
- What it does and its limits. Short and plain: what the system is for, what it cannot do, and when a person checks its work.
- How to reach a human. A clear route, not a maze.
- How to contest a decision. If AI shapes a decision about a person, they should know how to ask for a review.
Region notes
These are summaries. Always check the region page and the primary source.
- EU: The AI Act’s Article 50 transparency duties apply from 2 August 2026, including machine-readable marking of AI-generated content. Systems already on the market before then have until 2 December 2026 for the marking duty.
- US: Varies by state, for example disclosure of generative AI interactions in Utah and notice to employees and applicants in Illinois.
- UK: Through existing law and regulator guidance, and safeguards for significant automated decisions. Commencement of reforms is not settled.
- India: Intermediaries must label synthetic content and attach provenance metadata under the 2026 IT Rules amendment.
Inward transparency: data lineage
A system is only as explainable as its records. Keep track of where training data came from (origin, licence and collection conditions), how it was cleaned and transformed, and which version was used. A common audit gap is plenty of logs but no record of the dataset version behind a model.
Example
A retailer’s chatbot answered refund questions but never said it was automated, and gave no way to reach a person. After complaints the retailer added a clear “you are talking to an AI assistant” line, a one-click route to a human and a short note on what the assistant cannot decide.
What to do now
- List where your customers or staff meet AI. Is there a notice at each point?
- Use the AI transparency notice template when it is available in a later phase, and check the EU page for dates.
Sources
- Supplied governance study notes (module 2)
Reviewed 2 Oct 2026. Sources: EU AI Act, Regulation (EU) 2024/1689; MeitY (Ministry of Electronics and Information Technology).
Terms used: Provenance metadata, Synthetic content, Training data, Transparency.
Part of: Everyone path, HR path.
Was this page useful? Tell us what is missing or wrong ↗ · Report a correction