Your Agency Uses AI. Do You Know How? Closing the Agency AI Transparency Gap
When your own team uses AI to write copy or make images, you have a fair idea of what is going on. The work that comes back from your ad agency, design studio or social media freelancer is another matter. How much of it was made with AI, with which tools, and with what data fed in? Research published by the World Federation of Advertisers (WFA) on 30 September 2026 gives a blunt answer: most brands do not know. Agency AI transparency is becoming the most overlooked issue in marketing partnerships, and the one most worth settling first.
The agency AI transparency gap: 96% versus 13%
The WFA surveyed 54 senior respondents at 46 major brand owners with a combined annual ad spend of about $93 billion. The key numbers:
- 96% of brands now use generative or agentic AI in marketing, up from 45% in 2023.
- 80% use it for content creation, 72% for research and insights, and 54% for media buying and optimisation.
- Only 13% say they have substantial visibility into how their agencies use AI.
- 69% require agencies to disclose AI usage and 59% have added AI-related clauses to contracts.
Even the best-funded brands are asking for disclosure without yet seeing clearly. For smaller brands with no legal team, the gap is wider still.
Where the risk hides
The trouble with not knowing how an agency uses AI has little to do with AI being used at all. It comes down to four things:
- Rights in the assets. Are ownership and licence terms clear for AI-generated images, video and music? Can the work be registered and used commercially for the long term?
- Confidential data. Has your customer list or unreleased product information been pasted into a public AI tool?
- Brand consistency. AI output with nobody checking it lets tone and visuals drift.
- Disclosure duties. Rules on labelling AI-generated content keep multiplying, and when something goes wrong the brand is named, not the agency.
Marketing Dive reported in August 2026 that the IAB has released version two of its AI Transparency and Disclosure Framework, taking in new rules in the EU, Asia, New York and California. Data cited in the report shows 83% of ad executives have used AI in the creative process and 72% of marketers believe AI disclosure needs an industry standard. For the consumer side of the story, see our piece on the AI ad trust gap.
Five things to settle in the contract
Writing in O'Dwyer's in September 2026, two lawyers from Davis+Gilbert advised agencies and clients to address AI directly in the master service agreement or an amendment. From the brand's side, five points matter:
- Proactive disclosure. Which work will involve AI is discussed up front, before anyone discovers it later.
- A tiered approach. Routine internal use such as meeting notes and scheduling is treated separately from public-facing deliverables, and specific uses can be approved in advance.
- Clear responsibility. Who reviews the work and who bears the risk when it goes wrong is written down.
- Flow-down. Subcontractors and freelancers hired by the agency follow the same rules.
- Governance. An approved-tools list, escalation for higher-risk uses, and staff training.
The WFA research shows brands doing their own homework too: 75% require human oversight of AI outputs, and 55% now have centrally approved tool lists, up from 18% in 2023.
Transparency is not a ban
Asking for transparency does not mean telling agencies to stop using AI. A ban only pushes the use out of sight, and the brand sees even less. The IAB's updated framework also warns about label fatigue: routine editing and stylised imagery do not need disclosure, and labelling everything makes the label meaningless.
A more practical split looks like this:
- AI handles first-draft proposals, asset variations, resizing and data tidying.
- People handle brand judgment, final review, the decision to publish, and accountability for the result.
A good agency is comfortable being asked. Being able to say where AI was used and where a person did the work is part of being professional.
Three questions any small brand can ask
You can start without a legal team:
- Which parts of this project will use AI, and which tools?
- Will the data I provide be entered into those tools, and is it used to train models?
- Who does the final review of the deliverables, and what happens if a rights dispute comes up over an asset?
Put the answers in the quote or a short memo and you are already ahead of most brands. At TitApeX we say these things up front. AI handles perfection, people create what moves you, and clients have a right to know who did which part. To see how we use AI in projects, browse our services or get in touch.
Comments