The data on this is more dramatic than most marketers realize. Survey research this year found a large majority of consumers say AI makes ads feel less authentic, and a majority are less likely to buy from brands that visibly lead with AI-generated content. Separately, advertisers have been shown to significantly overestimate how comfortable younger consumers actually are with AI-generated marketing.
McDonald’s Netherlands pulled an AI-generated holiday ad this year after public backlash calling it soulless. Coca-Cola faced similar criticism despite a significant production investment. The pattern across both is consistent: AI showed up in a context where the audience was specifically looking for human warmth, and they felt its absence.
There are two distinct backlashes, and they require different responses
The “AI slop” reaction is fatigue with low-effort, mass-produced synthetic content: generic captions, uncanny imagery, templated copy that hits every expected beat without saying anything interesting. This is a quality problem, and it is fixable with better process and a human review gate.
The “AI-first brand” reaction is narrower and more strategic: audiences rejecting brands that make AI their identity, that announce their use of AI like it is the innovation itself rather than infrastructure behind the scenes. This is a positioning problem, not a quality problem, and no amount of better AI output fixes it.
What the brands succeeding with AI actually do
They use it behind the scenes to work faster, test more creative variations, personalize at scale, and optimize performance, and they do not announce it. AI becomes infrastructure, the way marketing automation software or analytics tooling is infrastructure, not a headline feature of the brand story.
Practical guardrails worth adopting:
- Enforce one consistent human brand voice across every AI-assisted output, rather than shipping whatever tone the model defaults to.
- Keep a human review gate on anything customer-facing, no exceptions, even when the AI-drafted version looks ready to ship.
- Reserve AI-first messaging for contexts where efficiency genuinely is the value proposition (a dev tool, an internal ops product), and avoid it entirely in categories built on craft, trust, or emotional connection.
- Lean on real customer photos, named team members, and founder visibility as a deliberate counterweight, especially in the parts of your marketing where authenticity carries the most weight.
The backlash is not a reason to slow down AI adoption in marketing operations. It is a reason to keep AI where it belongs, in the workflow, and keep the brand voice where it belongs, sounding like a human who actually cares about the customer.
Why McDonald’s and Coca-Cola are the instructive cases, not the outliers
It’s worth being specific about what actually went wrong in those two campaigns, because the lesson isn’t “avoid AI in advertising” — plenty of AI-assisted work from both companies and their competitors has shipped without incident. The specific failure was context mismatch: a holiday ad and a heritage-brand campaign are both categories where the entire value proposition is emotional warmth and craft, the exact qualities audiences are primed to detect as absent in AI-generated output, whether or not that detection is fully accurate. The same underlying AI tooling used to generate rapid creative variations for a performance-marketing test, where the audience’s expectation is efficiency and relevance rather than emotional authenticity, wouldn’t trigger the same reaction. The lesson generalizes: the risk isn’t the tool, it’s using the tool in a category where the audience’s implicit expectation is specifically the thing AI is worst at signaling — genuine, considered human craft.
The overestimation data point is the more useful one for planning
The finding that advertisers overestimate younger consumers’ comfort with AI-generated marketing deserves more attention than it usually gets, because it’s the data point that actually explains why brands keep making this mistake despite the backlash being well documented. Marketing teams skew younger and more online than the general population, and teams that work with AI tools daily tend to normalize AI-generated output faster than their actual audience does — familiarity breeds comfort for the people doing the making, not necessarily for the people on the receiving end. That gap is worth building a specific check into your creative review process: before anything AI-assisted and customer-facing ships, have someone outside the immediate creative team — ideally someone closer to the actual target audience’s relationship with the brand — react to it cold, without knowing AI was involved in producing it. If their gut reaction is different from the team’s, that’s the signal the internal team’s calibration has drifted from the audience’s.