Tell customers when AI helped draft your content. In a 2026 Gallup survey, drafting help earned wide acceptance once it was disclosed, while AI-made people and voices drew rejection even with a label. Where AI sits in your content marketing decides how buyers react.
The law sits well behind buyer attitudes. The FTC’s reviews and testimonials rule sets no AI-disclosure duty and has no blanket ban on AI-generated avatars. Even so, 62% of adults in Gallup’s survey called AI-made people or voices unacceptable, and that held when the business clearly disclosed the use.
If your vendor or staff already uses AI on posts and images, a single question comes first. What job is the AI doing?
Buyers Sort AI Content by the Job It Does
Gallup’s 2026 survey split business AI use into three jobs, and respondents graded each one differently. The Bentley University-Gallup Business in Society study asked 3,270 U.S. adults about advertising from May 4 to 11, 2026. Blog and social posts weren’t part of the question, so your own content may land differently.
The shares calling each approach acceptable differed sharply by job:
| What the AI does | Acceptable | Condition in the survey |
|---|---|---|
| Brainstorming or early drafts | 75% | Clearly disclosed |
| Final text, images, or videos | 53% | None stated |
| Creating people or voices | 38% | Clearly disclosed |
Overall, 49% of respondents viewed businesses’ AI use in ads negatively, while 19% viewed it positively and 32% stayed neutral. Each task asks something different of the reader, so a blanket policy for all AI work ignores differences the survey shows.
Most Americans Accept Disclosed AI Help With Drafts
Three in four respondents, 75%, called AI-assisted brainstorming and early drafts acceptable when the business is upfront about it. That question carried a clear-disclosure condition, and because Gallup didn’t report results for undisclosed work, the figure says nothing about what silence costs. It shows acceptance when the audience is told.
Drafting earned the highest acceptance of the three jobs, which makes it the easiest place to start telling customers. Where human review of AI drafts is part of the workflow, the notice can say so, but only if someone does the review every time.
Acceptance Drops When AI Finishes the Ad Itself
A smaller majority, 53%, accepts AI-produced final text, images, or videos, while another 46% call that approach unacceptable. Gallup attached no disclosure condition to this question. The drop from 75% therefore mixes a bigger AI role with no stated notice, and the data can’t separate them.
Age splits the answer sharply. Among adults 18 to 29, 60% call AI-finished content unacceptable, compared with 35% of those 45 to 59. If your vendor produces AI-generated campaign assets, you’re working in this middle ground. Nearly as many respondents objected as approved.
Most Americans Reject AI-Made People and Voices Even With a Label
In Gallup’s survey, 62% said it’s unacceptable for a brand to create people or voices with AI. That view held even when it clearly disclosed the use, and the other 38% found it acceptable.
A label tells viewers what happened, and most of them still object. Among adults 18 to 29, objection runs highest, at 73%. The category covers voices as well as faces, so an AI voiceover sits on the same side of the line as a synthetic spokesperson. For an owner weighing either one, that 62% is the cost to price in, since a label alone doesn’t remove it.
The Law Asks Little While Buyers Ask for More
The FTC’s reviews and testimonials rule sets no duty to label AI-produced content. New York’s synthetic performer statute covers a narrow slice of advertising. Buyers draw the line elsewhere. Neither bans AI-made people, yet most of Gallup’s respondents reject them even with a label.
Federal Rules Focus on Fake Reviews and Endorsements
The FTC’s final rule reaches testimonials and reviews that misrepresent who wrote them, “such as AI-generated fake reviews.” It lets the agency seek civil penalties from knowing violators. Anyone who buys or spreads such reviews faces liability when they knew or should have known the content was false.
For everything else, the staff answers say the rule has “no blanket prohibition” on AI-generated avatars in marketing. Its only disclosure requirements involve company insiders. A testimonial delivered by an AI avatar stays prohibited when the testimonial itself is fake or false, and the FTC adds that such avatars could also be deceptive under the FTC Act. The FTC calls those answers neither definitive nor a safe harbor. Questions about your own ads belong with an attorney.
New York Requires a Label on AI-Made Performers in Ads
Under General Business Law 396-b, a business that creates an ad has to disclose a synthetic performer conspicuously. A synthetic performer is a digitally created figure meant to give the impression of a human performer. The statute exempts audio ads and translation-only AI. It applies when the advertiser has actual knowledge that such a figure appears.
Penalties run $1,000 the first time and $5,000 after that. Whether the statute reaches a company based outside New York isn’t confirmed. Neither it nor the FTC rule covers AI-drafted text, which leaves what to tell customers about that material largely up to you.
Match What You Tell Customers to What the AI Did
Acceptance falls as the AI moves from drafting to finishing to creating people. The figures run 75%, 53%, and 38% in that order. The same numbers produce a working rule of thumb, sorted by the AI’s job:
| What the AI did | What Gallup found | What to tell customers |
|---|---|---|
| Drafts and brainstorming | 75% accept when disclosed | Say so plainly where the content appears |
| Final text, images, or video | 53% accept, 46% don’t | Disclose, and expect a split reaction |
| People or voices | 62% reject even when disclosed | Decide whether to use them before deciding how to label them |
Whether to Tell Customers
Gallup tested a clear disclosure only on drafting help, and 75% accepted it under that condition. For AI-finished content the survey set no condition, so silence carries a risk it can’t size, and for AI-made people or voices a notice leaves most objections in place. The first decision there is whether to use them at all.
Where to Put the Disclosure
Place the notice where the AI’s output appears. A note on a separate page doesn’t reach a reader who only sees the post itself. The FTC’s insider-disclosure requirement and New York’s statute both call for a conspicuous disclosure. The FTC treats a disclosure as avoidable when a reader has to click or hover to find it.
Though they cover other situations, those standards make a sensible model for an AI notice too. That means a caption or a line in the post. For an image, a caption directly beneath it does the same job.
How to Word It
A label should name the AI’s task in plain terms, using wording a customer would say out loud. A vague line like “made with modern tools” tells a customer nothing about what the AI did. These sample lines cover each task:
- “We used AI to brainstorm and draft this, and a person reviewed it before it went live.”
- “This image was made with AI.”
- “The spokesperson in this video is AI-generated.”
Keep each sample accurate, and drop the review clause from the first when no person checked the draft.
Write Your AI Line Down Before a Customer Asks
One short document settles this decision. It lists what your vendor or staff uses AI for and the wording customers see for each task. With it on paper, any team member can give the same answer when someone asks.
Our team is glad to talk through your situation with you. We can start by sorting out where AI already touches your content.













