Should agencies tell clients when they use AI?
Usually yes. Telling clients plainly how you use AI builds trust. What their audiences need to be told is a separate question.
Two audiences, two questions
Disclosure in agency work splits into two decisions. The first is what you tell your client about how you produced their work. The second is what the client's customers, followers or attendees need to be told about the content they see. Mixing them up leads either to oversharing that confuses, or to silence where it matters.
Telling clients
There is no blanket UK legal requirement to disclose the use of AI in advertising. However, client contracts may require disclosure, so check your terms and any supplier codes of conduct the client has asked you to sign.
Even where nothing requires it, proactive disclosure usually works in your favour. Clients increasingly assume agencies use AI. What damages relationships is discovering it later, for example when a PR client spots near-identical phrasing across two press releases, or a brand manager learns that a product shot they assumed was photographed was generated.
How to disclose without undermining the work
Describe the process, not the tool: "Our copywriters use AI writing assistants for research summaries and first-draft variations, then write and edit the final copy" tells the client what matters. Brand names and model versions add little.
Lead with human ownership: Make clear who sets strategy, who makes creative decisions and who signs off. You sell judgement and craft, and disclosure should reinforce that.
Keep it matter-of-fact: A short section in your credentials deck, statement of work or onboarding pack works better than a nervous confession halfway through a project.
Say where you don't use AI: If you keep AI away from brand identity, sensitive client data or final approvals, saying so reassures clients.
Offer an opt-out: Some clients, particularly in regulated sectors, will want AI kept out of certain work. Asking early avoids rework.
How you disclose also shapes your fee conversations, covered in how to price agency work when AI makes it faster.
Telling audiences
The UK Advertising Codes apply to ads regardless of how they were created, including where AI was used. The test is whether the ad materially misleads, not which tool made it.
That means a label does not fix a problem. Disclosing AI use does not cure an ad that is otherwise misleading. An AI-generated image of a skincare result that exaggerates what the product does stays misleading even with an "AI-generated" tag.
A label is useful where it helps audiences understand what they are seeing, such as a synthetic visualisation of an event venue or an illustrated scenario in a video ad. The UK government has said it will work with industry on best practice for labelling AI-generated content, so expect expectations to firm up.
Work aimed at EU audiences
If a campaign targets people in the EU, the EU AI Act may apply. Article 50 transparency obligations apply from 2 August 2026, including disclosure obligations for certain AI-generated or manipulated content. An events agency running a product launch across European markets, or a social team managing an EU-facing account, should check whether its content falls within scope.
Put it in writing
Agree disclosure expectations with each client and record them, ideally in the contract itself. Write down what you will tell the client, what the client will tell its audiences, and who is responsible for any labels. Advertising and contract rules depend on the specifics, so take advice on your own situation.
More questions