Is your AI generated content weakening trust?
This is a question I am beginning to ask more. Not “can AI do this?” but “should we be using it here, and what does it say about us if we do?”
I had a very specific version of this on a recent project. We needed imagery of real people using a product in real-world situations and we didn’t have enough photos to pull from. It’s not an unusual problem. Budgets are tight, timelines are always shorter than you’d like, and AI image generation felt like an option. So, we tried it.
When AI felt like the obvious answer
Once we’d determined the best prompts, the images outputs were very good. The product looked good, the overall image was on brand. The environments looked real enough and I retouched over some sections that did look a bit questionable. We sat with them for a bit and started asking ourselves about whether we felt comfortable using them.
Then we started asking questions
The first question was: where did these faces come from? AI image generators are trained on data. Real people are in that data. We had no idea who those faces belonged to, whether they’d consented to being used, or whether we were about to put someone’s likeness on a marketing asset without any right to. That didn’t feel right at all.
The second concern was simpler but just as serious. We were generating images of people using the product in a way that implied a real experience. But those weren’t real users. Nobody had actually sat with the product. We were creating social proof that didn’t exist.
Put those two things together, faces we couldn’t account for, and users we’d invented, and we saw exactly where it could lead. In a world where buyers are already suspicious of what’s real and what isn’t, and everyone is in a game of ‘Spot the AI’ right now, and that didn’t feel comfortable at all.
So we scrapped them all and found a different solution. One that used a bit less AI, took a little longer and was more considered. The result felt much more honest about what we were actually trying to communicate. And more importantly, it was something we could stand behind without any caveats.
That’s what AI-generated content is on danger of getting wrong when it’s used in a rush. The output might look ok, but the questions behind it matter just as much. Who is this? Is this real? Does this represent what we’re actually selling? Those are trust questions. And trust, once someone feels misled, is waaaay harder to rebuild than it is to protect in the first place.
The questions matter as much as the output
This isn’t about AI being good or bad. I use these tools and find them useful. But there’s a difference between using AI to work faster and smarter and using it to fill a gap in a way that introduces risk you haven’t thought through properly.
For businesses in sectors where trust is effectively the product, specifically for users of Scail’s products, regulated SaaS sits squarely in that category, this matters at a completely different level. Buyers in those markets are already on high alert. They’re not just browsing your website. They’re checking whether what you’re saying adds up, looking for evidence of care and control at every touchpoint. AI-generated imagery that doesn’t represent real users, or content that implies a capability the product hasn’t earned yet, doesn’t just create friction. It can create doubt that unravels a lot of hard work very fast.
The question worth asking before you use any AI-generated content isn’t just “does this look good?” It’s “does this feel true? And what happens to trust if it doesn’t?”
If you want to understand where your AI output reads well from the outside and where it might be creating gaps in trust without you noticing, Scail’s Risk & Value Scorecard looks at the full picture, including the visible-trust gaps that are easiest to miss when you’re inside the work.
Read more about our AI Risk & Value Scorecard.
