On 20 October, Cint and Advertising Week published The Authenticity Audit, a survey of around 1,000 US consumers. Three in four said brands have a duty to disclose when AI has been used in advertising and marketing. Nearly half dislike seeing AI-generated people in ads at all. If you run a brand, AI disclosure in marketing has just moved from a philosophical debate to a practical question with a deadline.
This is not an anti-AI piece. We use these tools at SOHO every week, and so do most of our clients. Instead, this article sets out what audiences actually object to and where AI tends to sit in a creative process. It then shows how to write a disclosure policy your team can follow without a lawyer on speed dial.
Why is AI disclosure in marketing suddenly on the agenda?
The Cint numbers are blunt. Alongside the 74 per cent who want disclosure, 44 per cent said that discovering undisclosed AI advertising would lower their trust in a brand. Undisclosed AI or synthetic media was the top reason for losing trust, ahead of dodgy influencer deals and disguised sponsored content. Meanwhile, 93 per cent said trust affects what they buy.
The mood on the conference floor matches the data. Reporting from Advertising Week New York on 21 October, Columbia News Service found marketers discussing AI’s limits as much as its promise. One research executive noted more public scepticism than a year earlier. Then, in Creative Review on 22 October, Ellen Utrecht argued that AI has exposed cracks in the creative industry that were already there. She listed pressure for volume, processes that flatten ideas, and too little room for judgement. Blaming the technology for everything, she wrote, “lets the creative industry off the hook”.
Put those together and the picture is clear. Audiences are not rejecting technology. They are rejecting the feeling of being fooled, and the sense that nobody cared enough to make the thing properly.
What do people actually object to?
Here the detail matters, because it tells you where to be careful. In the Cint study, AI-generated people were the most disliked element, at 46 per cent. Realistic landscapes were the most liked, at 40 per cent. In other words, a generated sky behind your product is a shrug. A generated customer smiling at your product is a problem.
Two forces are at work. The first is deception: people have no way to check whether that face is real, so the whole message starts to feel like a trick. Effort is the second. A campaign that looks like it took an afternoon tells the audience how much the brand values them. Premium brands feel this most sharply, because effort is a large part of what they sell.
For the companies we work with in fintech, deep tech and health technology, such as Simudyne and Withings, the stakes are similar. Their customers are buying trust in a complex product, often before they fully understand it. Anything that reads as synthetic makes that leap harder.
Where does AI already sit in your creative process?
AI disclosure in marketing starts with an honest map. In most agencies and marketing teams, AI shows up in three layers, and they deserve very different treatment.
The thinking layer
Research summaries, competitor audits, the first draft of a brief, a messy brainstorm at 11pm. Nobody expects a footnote here, any more than they expect you to credit the search engine.
The production layer
Retouching, background extension, layout variants, translation drafts, resizing for forty social formats. This is where most of the time savings live. It is also where a stray generated hand can slip through unnoticed.
The front-of-house layer
Hero imagery, people, voices, testimonials, product claims and anything presented as real. This is the layer the Cint respondents care about. Consequently, it is where disclosure, or a decision not to use AI at all, belongs.
Most disclosure debates go wrong because they treat all three layers as one thing. They are not.
How do you write an AI disclosure policy people will follow?
Here is the process we use when a client asks for help with AI disclosure in marketing. It fits on two pages and takes an afternoon, provided the right people are in the room.
1. Draw three lines
Decide what is never generated: customer faces, employee portraits, product shots that claim accuracy, testimonials. Next, list what is generated with review: backgrounds, textures, concept visuals, social variants. Finally, name what is free to use: research, drafting, ideation. Write it all down with examples.
2. Choose your disclosure levels
Not everything needs a label. Internal logging covers the thinking layer. Disclosure on request suits production work. A visible label belongs on anything synthetic that could be mistaken for real. In short, match the level to the risk of being misread.
3. Agree the wording and placement
Avoid legalese. “Visuals created with AI tools and reviewed by our team” tells people the truth in ten words. Put it where the content lives, not buried in a terms page.
4. Keep a record
Note which tools were used, who approved the output and whether any rights-sensitive material was involved. If a journalist or regulator asks in eighteen months, you want the answer in a spreadsheet, not in someone’s memory.
5. Brief every supplier
Your policy only works if your agency, freelancers and media partners follow it. So put it in the brief and in the contract. We would much rather be held to a client’s policy than guess at it.
Which mistakes should you avoid?
The first mistake in AI disclosure in marketing is over-disclaiming. Stamping “made with AI” on a spell-checked caption trains people to ignore the label when it matters. The second is the hollow pledge. Declaring your brand “AI-free” while your team uses AI daily will eventually surface, and the Cint data suggests undisclosed use costs more trust than disclosed use.
The third mistake is treating disclosure as a legal exercise only. Regulation is arriving, including transparency duties under the EU AI Act for certain synthetic content, and your legal team should track it. However, the audience is already ahead of the law. They want to know whether you respect them enough to be honest with them.
Finally, do not let the debate stall your work. The Creative Review piece is right that AI did not create the industry’s taste for speed over originality. Disclosure is simply a chance to show which side of that line your brand sits on.
A conversation, not a sales pitch
If you are working out where AI belongs in your brand, and how AI disclosure in marketing should work for you, we are happy to compare notes. Our brand strategy work usually starts with exactly these questions, and we have written before about choosing between AI imagery and real photography. Get in touch and tell us what you are wrestling with.
Frequently asked questions
Is AI disclosure in marketing a legal requirement?
It depends on your market and the content type. The EU AI Act introduces transparency obligations for certain AI-generated or manipulated content, and advertising regulators in several countries already act against misleading imagery. Treat the Cint figures as the audience’s standard, and check the legal baseline with counsel.
Does AI disclosure in marketing hurt campaign performance?
The evidence so far points the other way for trust. In the Cint survey, undisclosed AI was the top reason people lose faith in a brand. The dislike of AI-generated people was about the content itself, not the label. A clear, calm note tends to defuse suspicion rather than create it.
What about AI used only for research and ideation?
Log it internally and move on. Audiences are concerned about what they see and whether it is real, not about how you organised your notes. Save visible disclosure for synthetic content that could be mistaken for the real thing.
