Too much AI, too little trust: what 2026’s consumer data asks of brands

The number that should worry marketers

In September 2026 Gartner published a survey of 1,006 US consumers, carried out in May and June. Sixty-five per cent of them said brands are producing too much AI-generated content in their marketing. Fifty-seven per cent said that content has made them less trusting of brand messaging overall. And 35% said AI has made them rely less on influencers for shopping information.

Put those figures side by side and a pattern appears. People aren’t objecting to a technology. They are objecting to volume, to sameness, and to the suspicion that nobody in particular is speaking. A feed full of glossy, slightly-off product shots and captions that could belong to any brand teaches people to scroll faster. It also teaches them to doubt the message, which is the expensive part.

In June the same firm reported that 49% of US consumers felt generative AI had made content quality worse. Two surveys are not a law of nature, and both are American. But nothing in Indian feeds suggests audiences here are more forgiving. A market where people compare prices on three apps before buying a phone charger is a market that notices when it is being worked.

Why “too much” is the phrase that matters

There is a useful distinction hiding in that 65%. The complaint is about quantity. Most people have used an AI tool themselves by now and know what it is good for. What they resent is a brand that seems to have handed its voice over to one.

We think it comes down to three things that AI-heavy marketing tends to lose:

• Specificity. Generated copy drifts towards the average of everything it has read. “Elevate your everyday” fits a mattress, a bank and a spice blend equally well, which means it says nothing about any of them.
• Evidence. A rendered kitchen that doesn’t exist can’t prove your appliance works in a real one. When the picture is invented, the claim starts to feel invented too.
• Someone answerable. Trust is partly a judgement about whether a person will stand behind what was said. Content that reads as nobody’s makes that judgement harder.

None of this is new to good marketing. Peter Drucker wrote in 1973 that the aim of marketing is to know and understand the customer so well that the product sells itself. What has changed is how cheap it has become to skip that step and still fill a content calendar.

India now has rules, and they point the same way

On 10 February 2026 the Ministry of Electronics and Information Technology notified amendments to the IT (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, which came into force on 20 February. They define “synthetically generated information”: audio, visual or audio-visual content created or altered by a computer so that it appears real, showing people or events in a way that could pass for the real thing. Platforms must label it, visibly and in machine-readable metadata, and must act on flagged unlawful content within three hours.

The duties fall mainly on platforms rather than advertisers, and legal summaries differ on the finer points, so read the notified text with your lawyer before you change a workflow. The direction is still plain. A realistic synthetic spokesperson, or a generated “customer” giving a testimonial, is exactly what these rules were written for. If a campaign depends on people not noticing that a face was generated, it is the wrong campaign for 2026.

Our own rule of thumb is short: if a reasonable viewer would feel misled on learning that something was generated, either label it or don’t make it.

Where AI earns its place

We use AI tools in the studio every week. Pretending otherwise would be silly, and clients who ask are usually relieved by a straight answer. The test we apply to each use is whether the customer meets the output directly, and whether a person has checked it.

Uses that tend to pass:

• Research and synthesis: summarising interviews, grouping thousands of reviews by theme, finding the questions customers keep asking. Someone still reads the source material.
• Production chores: cutting one master film into a dozen formats, cleaning up audio, rough storyboards that an illustrator then redraws.
• Options for testing: twenty headline routes so a writer can find the two worth sharpening.
• Media planning: reading demand signals faster than a weekly report can.

Uses that usually fail:

• Final brand copy that nobody has rewritten in the brand’s own voice.
• Product images showing things the product can’t do, or places it has never been.
• Synthetic people presented as real customers, employees or experts.
• Content made to fill a calendar rather than to say something.

The line isn’t “AI bad, human good”. A careless writer produces generic copy too. The line is accountability: everything that reaches a customer should have someone who can explain why it says what it says.

What to put in front instead

If AI takes on more of the hidden work, the visible work should become more human, not less. In practice that means a handful of unglamorous habits.

Show the real thing

Photograph the product, the factory floor, the founder who still answers support emails. When trust is the goal, a well-lit photo of a real object beats a flawless render of an imagined one. Where we use CGI, and our Production team uses it a great deal, it is to show something true more clearly: an exploded view of a device, a material at a scale no camera can reach.

Let people speak for you

Gartner’s 35% describes influencers losing ground because of AI, not people losing ground. Reviews, customer stories and creators who actually use the product still carry weight, possibly more than before. Choose fewer creators, choose them for fit, and let them speak in their own words. That is where our PR & Creator work starts.

Make claims you can back up

A believable brand message has a source. If you say “loved by 10,000 customers”, know where the number comes from. If a page explains how your product works, have the engineer read it. Search engines and AI assistants increasingly reward this too, because they are trying to judge the same thing your customer is.

Disclose without drama

A short line such as “Background created with AI; product photographed in studio” costs nothing and heads off an awkward discovery later. Disclosure works best when it is matter-of-fact. It reads as confidence.

A quick audit you can run this week

Pull your last thirty published pieces and ask four questions of each:

1. Could a competitor publish this with their logo swapped in? If so, it is too generic.
2. Does it show anything real: a product, a person, a place, a number with a source?
3. If any part was generated, would a viewer feel misled on finding out?
4. Is there a named team or person who would stand behind it?

Expect a fair share to fail the first question. That isn’t a disaster. It is a list of things to stop doing, which frees budget and attention for the few pieces that deserve real care.

The opportunity in all this

When anyone can produce content for almost nothing, content stops being scarce. Judgement, taste, proof and a recognisable voice become scarce instead. Those were always the hard parts of marketing, and they are the parts AI can assist but not replace.

That is good news for brands prepared to do the work. Because so much of what is published looks alike, the bar for standing out has dropped. A clear point of view, shown with real evidence, is now unusual enough to be noticed.

If you would like a second opinion on where AI fits in your marketing, and where it doesn’t, tell us what you’re working on.