The AI Trust Gap Insight: The One Distinction That Separates AI Use That Builds Trust From AI Use That Erodes It
Author: IMB Editorial Team
IMB Journal – International Marketing Board
Volume 1 | Issue 6
June 2026
The One Distinction That Separates AI Use That Builds Trust From AI Use That Erodes It
Every brand debating how far to take AI in its marketing is really debating two different questions at once, and most of the public missteps this year have come from answering them as if they were the same question.
The first question is where AI produces something the customer directly sees or hears: the copy in an ad, the footage in a commercial, the voice reading a script, the face on screen. Call this the generative use of AI. The second is where AI shapes decisions the customer never observes directly: which audience segment sees which message, what time an email goes out, how a media budget gets reallocated between channels overnight, which product gets recommended next. Call this the operational use of AI.
The backlash documented throughout this year, from McDonald’s Netherlands to Coca-Cola’s holiday campaigns, has landed almost entirely on the first category. Nobody has organized a social media pile-on because a brand’s ad-targeting algorithm got more efficient. The criticism shows up precisely when AI moves from deciding what happens behind the scenes to visibly standing in for a human voice, face, or creative choice the audience assumed was made by a person.
This gives marketing teams a genuinely simple test to apply before greenlighting any new use of AI: will the customer ever see or hear the specific output this system produces? If the answer is no, the operational category, the efficiency case for using AI aggressively is strong and the reputational risk is close to zero. If the answer is yes, the generative category, the same efficiency case still exists, but it now has to be weighed against a real and currently rising probability that the audience will notice, and that noticing will read as a small deception rather than a neutral production choice.
This doesn’t mean generative AI is off-limits in customer-facing work. Plenty of brands use it well in early drafts, storyboarding, or as a starting point that a human team substantially reworks before anything ships. The distinction that matters isn’t whether AI touched the project at all. It’s whether the final, customer-facing output is something AI produced or something AI merely assisted a person in producing, and whether the brand is honest about which of those happened if anyone asks.
Applying this test doesn’t require a new department or a policy binder. It requires one question, asked before a campaign goes live rather than after it draws the wrong kind of attention: are we asking AI to do the parts of this job our audience will never see, or the parts they came here specifically to experience as human?
This concludes our three-part series on AI and advertising trust. Read Part 1, “Why Brands Keep Getting AI-Generated Advertising Wrong,” and Part 2, the McDonald’s Netherlands case study, in this issue.