Pulled in Days: McDonald's, Valentino and Why "No AI" Is Becoming a Selling Point

In December 2025 two major brands pulled AI-generated ads within days of posting them. Two months later, a vodka brand ran a mostly AI-generated spot in the Super Bowl. Since then, a growing list of brands has started advertising that it didn't use AI. Our view is that "human-made" is becoming a premium production feature you can sell. AI in emotional brand work now carries a reputational risk, and that risk belongs in the treatment, not just the budget.

The two ads that came down

McDonald's Netherlands. TBWA/Neboko and the production company The Sweetshop made a 45-second Christmas spot built around "the most terrible time of the year": presents falling off a car, Santa stuck in traffic, and an invitation to hide out in McDonald's until January. It went up on a Saturday and was withdrawn by the following Wednesday after a social media backlash (NBC News). McDonald's disabled comments before taking it down (TechRadar). The production defended the work. The Sweetshop's CEO said the team was "ten people, five weeks, full-time" and that the effort "far exceeded a traditional shoot". She described generating "thousands of takes" and shaping them in the edit.

Valentino. On 2 December 2025 the brand posted an AI video on Instagram for its DeVain handbag. In it, models emerge from the bag and the logo turns into twisting human arms. The post was labelled as AI. Commenters called it "cheap", "lazy" and "disturbing" (Glossy, The Star / AFP). A McGraw Hill blog post says Valentino withdrew the campaign. We could not confirm this from the brand or from the contemporaneous news reports we checked, so treat the withdrawal as reported, not confirmed.

These cases show that effort doesn't protect you and labelling doesn't protect you either. McDonald's had a full-time team on the job, and Valentino disclosed the AI. Audiences judged what was on screen and what it seemed to say about the brand.

The one that ran anyway

Svedka's "Shake Your Bots Off" aired during Super Bowl LX on 8 February 2026. It was made with Silverside AI, the studio behind Coca-Cola's controversial AI ads, and billed as the first "primarily" AI-generated national Super Bowl commercial. According to TechCrunch, humans wrote the storyline. AI rebuilt the brand's robot mascot and handled most of the production. The sources disagree on how long it took. TechCrunch says about four months, while ComfyUI's case study says under one month. Svedka's Amy Remley told Comfy that "we had just as many brains and hands touching this ad as any traditional production."

Reaction ranged from indifferent to hostile. Adweek's creative roundup was headlined "Svedka's Forgettable Super Bowl Ad Proves AI Isn't an Idea", and The Takeout called it "nightmare fuel". But it stayed on air. It was about robots, so the synthetic look fitted the subject. The AI was also the stunt itself, and nobody was asked to feel anything about Christmas or craftsmanship.

What the failures had in common

Three ads aren't a dataset, so take this as a pattern to watch for and not a rule. The ads that came down shared three traits:

• An emotional category. Christmas and luxury both depend on the idea that someone cared. AI imagery seems to undercut that more than it would in, say, a tech or utility ad.
• Human faces and bodies. Both pulled ads relied on people and gave them the morphing, slightly-wrong bodies that viewers now recognise straight away. Svedka's main characters are robots.
• A synthetic look that didn't fit the subject. In the Svedka spot the robot subject justified the look. In the two pulled ads nothing on screen explained it, so viewers read it as cost-cutting.

"No AI" as a selling point

Brands are now using the backlash in their own ads. Inc. reported in January 2026 on Almond Breeze using the Jonas Brothers to mock AI ads, and on Equinox setting AI fakes against real people under the line "Question everything but yourself". It also cited Polaroid billboards for analogue photography, Heineken outdoor ads, and Aerie's commitment to use only real people.

Christopher Nolan made a similar argument in July 2026. He told The Telegraph that young audiences are "utterly rejecting" generative AI content and that he had "never seen a technology get dismissed" so fast. We have seen these quotes only in a secondary report, because the Telegraph original is paywalled, so treat the wording as reported.

For production companies, this flips the usual pitch. A real shoot with real people used to be the default and didn't need selling. For brands that want to look as if they care, it is becoming something the client will pay for. You can only sell it if you can document it: behind-the-scenes stills, crew lists, a written statement of where AI was and wasn't used. We think that paperwork is worth building into every emotional-category job now.

A pre-pitch risk checklist

Before you put generative AI into a treatment for emotional brand work, run through this. If you answer "yes" to two or more, we would pitch a live-action or hybrid route, and use AI only for invisible work (cleanup, extensions, previs).

1. Is the category built on care or craftsmanship: holidays, luxury, food, family, health?
2. Will AI-generated human faces or bodies be on screen for more than a beat?
3. Is the AI in the press release rather than just in the pipeline?
4. Would the synthetic look seem unmotivated to a viewer, with no robot, dream or fantasy premise to explain it?
5. Has the brand built its reputation on human skill?
6. Is there no plan for comments, press questions or a quick takedown in the first 72 hours?

If the job still goes ahead with AI at the front, write the takedown scenario into the contract. In both December cases the ads were dead within days, and the production company had to defend the work in public. Make sure the client has agreed in writing to how that defence will be handled.