Most US adults do not trust what they read on social media. A January 2025 Statista survey found that 41% of adults in the United States said they did not trust information posted on social media platforms very often, and another 16% said they did not trust it at all. The brands leaning on AI-generated marketing are publishing into a room where the audience is skeptical of both the channel and the speaker.
The skepticism runs deeper than the feed. A 2025 study from the University of Melbourne’s Melbourne Business School and KPMG, covering 48,000 people across 47 countries, found that only 46% of people globally are willing to trust AI systems. So brands that publish AI-generated marketing are starting a campaign in a room where most of the room is already skeptical of both the channel and the speaker. The same dynamic is now playing out in financial services, where how AI is reshaping trust in banking has become a frontline customer-experience question.
The Trust Deficit Marketers Walk Into
The two trust gaps feed each other. The Statista figure captures a feed-level problem: readers have stopped treating social posts as authoritative by default. The KPMG figure captures a tool-level problem: readers have stopped treating AI output as authoritative by default.
A business that uses AI to fill its social feed is asking an audience that already distrusts the channel to trust content made by the tool they distrust the most. That combination pushes the trust cost onto the brand, not the platform. The 2025 brand trust report on consumer decisions found that 88% of consumers say trust is as important as price and quality when they choose a brand, so a trust penalty is a purchase-stage problem, not a vague reputational one.
What ‘AI Slop’ Actually Does to a Brand
The penalty is measurable. Researchers at the Nuremberg Institute for Market Decisions ran a representative study of 1,000 respondents in each of the US, the UK, and Germany, then two controlled experiments on AI-disclosed marketing. In the second experiment, the same six ads were shown to two groups, labeled either ‘AI-generated’ or ‘human-generated,’ and the AI-labeled versions were rated less useful and less natural.
The Nuremberg Institute study of AI marketing content called the effect a trust penalty: a bias where consumers react warily when they sense a message was created by a machine. The bias held even when the content was technically polished, and it showed up in the engagement data. Consumers were less inclined to research or buy the products when the ad was AI-labeled.
The findings reveal that employees use of AI at work is delivering performance benefits but also opening up risk from complacent and non-transparent use. They highlight the importance of effective governance and training, and creating a culture of responsible, open and accountable AI use.
Nicole Gillespie, who chairs the Trust research partnership at Melbourne Business School and led the 2025 KPMG-University of Melbourne study, put the same tension in workplace terms. On the consumer side, the same dynamic shows up: 20% of NIM’s respondents said they trust AI itself, and 21% said they trust AI companies and their promises. The study found 44% of consumers are aware that AI can create marketing content, and only 25% think they can reliably recognize it, so the audience that cannot tell the difference is also the audience that reacts most negatively when it finds out.
Coca-Cola’s 2023 attempt to remake its iconic holiday ad with generative AI offered an early look at the public version of this penalty. Viewers called the spot ‘soulless.’ The Nuremberg Institute cites the backlash as the kind of moment a brand walks into when it leads with AI in the creative seat. The audience’s perception of the process now sits in the product, even when AI makes a good ad. The efficiency case for AI content is real, and the trust cost of AI content is also real.
By the numbers:
- 46% of people globally are willing to trust AI systems (KPMG and University of Melbourne, 2025)
- 41% of US adults do not trust information on social media ‘very often’ (Statista, January 2025)
- 16% of US adults do not trust social media information at all (Statista, January 2025)
- 20% of consumers say they trust AI itself (Nuremberg Institute for Market Decisions)
- 88% of consumers say trust is as important as price and quality in brand choice (Edelman, 2025)
Three Strategies to Rebuild Trust Without Giving Up AI
The playbook is to use AI where it does not carry the brand voice, and to keep a real person visible where it does. The three moves below are the spine of a trust-first marketing strategy for small businesses that cannot afford to lose a customer over a content shortcut. Each one leans on a different mechanism.
Lead with human-led content
Verified customer testimonials, founder videos, employee-generated posts, and expert commentary are the content AI cannot fake. A face, a name, and a stake attach to the brand. A small-business owner who appears on camera once a week signals that a human is the one delivering the message. The reverse also holds. A brand whose visible content is all AI-generated avatars starts to look like a brand with no human in it.
Use AI as the assistant, not the author
AI is useful for the scaffolding audiences do not see: first drafts, research summaries, alt text, transcripts, social cuts from a longer video. It is risky for the sentences and images that carry the brand’s voice. The cleanest rule is to let AI write the working draft, then rewrite the final version in a human voice before it ships. The 2025 global study on trust in AI found 57% of employees say they hide their AI use and present AI-generated work as their own, and that same opacity is the dynamic a small business should avoid in its customer-facing marketing. The audience does not need to see every tool the brand used, but it does need to know a human was the editor.
Build a personal brand AI cannot impersonate
Thought leadership, the founder’s specific point of view on a specific niche, is one of the few content moats that scale without losing trust. Concrete formats include LinkedIn posts in the founder’s name, a monthly email newsletter, and a guest appearance on a relevant industry podcast. Each adds a layer of attribution AI cannot borrow. Over time the audience starts to recognize the voice before they read the byline, and the brand becomes harder to fake.
None of these moves require a large team or a large budget. They do require a real human on the record, a clear use of AI in the back room, and a point of view the founder is willing to defend. The same combination is rare in AI slop, and a small business can actually build it.
A Quick Audit for Your Next Post
Run the next piece of marketing through this short test before it goes live. If a post fails two or more of these checks, it is probably contributing to the trust problem instead of solving it. The test is short on purpose, so it can run before every send, not just at the quarterly content review.
- Is there a real person visible or credited? If not, who is the reader trusting?
- Could a reader tell, in five seconds, whether a human or a machine made this? If they cannot, neither can you.
- Does the post take a position, or does it hedge the way AI tends to?
- Is the AI use disclosed in a way that adds credibility, rather than reading as an apology?
- Will you still be comfortable with this post in two years, given that slop dates fast and a real point of view ages well?
Any post that fails two or more of these checks is feeding the trust problem. The fix is usually a single round of human editing and a name on the byline. The audit takes longer to read than to do.
Frequently Asked Questions
What is the AI content ‘trust penalty’?
It is the measurable drop in consumer trust and engagement that appears when audiences know, or suspect, that a piece of content was made by AI. The Nuremberg Institute for Market Decisions documented the effect in controlled experiments, where the same ads were rated less useful and less natural when labeled ‘AI-generated’ than when labeled ‘human-generated.’ The penalty shows up in clicks, research intent, and purchase willingness, not just vague reputation.
Why do consumers distrust AI-generated marketing?
Only 20% of consumers trust AI itself, and 21% trust AI companies and their promises, per a Nuremberg Institute study of 1,000 respondents each in the US, UK, and Germany. The same study found 44% of consumers are aware that AI can create marketing content, but only 25% think they can reliably recognize it. The audience that cannot tell the difference is also the audience that reacts most negatively when it finds out.
Does human-led content actually build more trust?
The Nuremberg Institute’s experiments found that identical ads labeled ‘human-generated’ were rated as more useful and more natural than the same ads labeled ‘AI-generated.’ Edelman’s 2025 Brand Trust Special Report found 88% of consumers treat trust as equal to price and quality when they choose a brand, so the lift from human-led content is a purchase consideration, not a feel-good metric.
How should a small business use AI in marketing without losing trust?
Keep AI on the tasks audiences do not see, like first drafts, research notes, transcripts, and alt text. Keep a real person, not an AI avatar, on the camera, the byline, and the customer-facing voice. The KPMG-Melbourne Business School study found 57% of employees say they hide their AI use at work, and that same opacity is the dynamic a small business should avoid in its marketing. Disclosure alone will not fix the trust deficit, per the Nuremberg Institute, so label AI use only when the label makes the content more credible.
What is ‘AI slop’?
The term refers to the flood of generic, low-effort AI-generated content that now fills social feeds: cloned voices, stock avatars, templated captions, recycled images. It is distinct from thoughtful AI-assisted content, and audiences can usually tell the difference. The risk for a small business is being grouped with the slop by default, which is why a visible human and a specific point of view are the easiest defenses to build.
