Can We Still Trust What We See Online?
I tried to tell real ads from AI-generated ones for a cosmetics brand, and I couldn’t
I’m an e-commerce fan. I spend a lot of my free time reading about online shops, ads and the strategies behind them. Recently, I watched a case study about an American cosmetics brand called Smooche. The video, “Le Shopify sorti de nulle part qui casse internet”, was made by the French creator Chris Navarro.
The author explains that Smooche makes about $750,000 in revenue per day, and that almost everything except the product itself would be AI-generated: images, videos and ads. On paper, it is an ordinary brand that sells makeup and skincare. What makes it different, according to him, is the way it is marketed.
I have no way of checking these numbers. But the idea stayed in my head, and it made me want to run a test. I opened the shop, then Meta’s Ad Library, where a search for “smooche” gives around 3,000 results. I scrolled through the ads and the product visuals, and I tried to guess which ones were real and which ones were not.
Honestly, I couldn’t tell. Some images looked like professional photo shoots. Some videos looked like a real customer talking to the camera. Maybe they were real. Maybe they were not. I had no way of knowing.
If I, who spend my time on these topics, can’t tell the difference, what about an ordinary customer who is just scrolling on their phone?
How AI Is Changing E-Commerce Marketing
Generative AI is now used in online selling in several places.
- Product photos: A brand can show a product in a setting, or in a person’s hand, without doing a photo shoot.
- Video ads: Short videos, like the content you see on social media, can be created by the dozen.
- Text: descriptions and ad slogans can be written and tested in minutes.
- Personalization: Emails, product recommendations and even homepage content can adapt to each shopper based on what they browse and buy.
- Customer service: AI chat assistants can answer questions about sizing, shipping or returns at any hour, without a support team on standby.
This is not classic automation. With a classic tool, you always said the same thing. Now you can say dozens of different things.
The Business Opportunity: Faster, Cheaper, More Creative
The advantages for a brand are clear:
The first one is price. A photo shoot or a film shoot needs a team, a place and equipment. The second one is speed: an idea can become an ad the same day. The third one is testing. A brand can test many versions of the same ad and keep the one that works best. And the last one is access for small businesses, which can now make visuals that look professional.
But there is a hidden cost. Someone has to check what comes out, fix the mistakes and make sure it looks like reality. The savings are real, but they are not always as big as they seem.
When AI Becomes a Problem for Consumer Trust
The first problem is that the line between a real photo and a generated one is getting thinner. A product shoot can look authentic when it isn’t. And for the customer, it is hard to detect. A video where a person recommends a product can look genuine when it is not. The product the customer receives may be different from the one they saw.
Get Rémi Rougetet’s stories in your inbox
Join Medium for free to get updates from this writer.
The second one is the lack of transparency. If the brand doesn’t say what is real and what isn’t, the customer has no way of knowing. And when they find out later, they start to wonder what else isn’t true.
A Real Example: Smooche
Smooche is a good example for this topic because everything is public. I will separate what I was able to see from what I cannot say.
What I could see:
The shop sells cosmetics: foundation, serum, and products for lips, lashes and brows. It shows impressive numbers: 1 million customers, more than 50,000 five-star reviews, and dermatologists who recommend the brand. In Meta’s Ad Library, many ads have several versions: the same idea, with slightly different text and images. Several announce a promotion that ends “tonight”, yet their start date is in June or September and they are still active. I don’t know if this is misleading, because the promotion may have been renewed. But it is the kind of detail that makes you doubt.
What the video says:
According to its author, almost everything except the product is made with AI. I didn’t see any AI mention on the shop’s homepage. Where Coca-Cola added a small note saying its ad was made with its own AI program, I saw nothing like it here. But I didn’t read the whole site, and the absence of a mention proves nothing.
What I cannot say:
I cannot prove which images or videos are real and which are simulated. The testimonial videos, where a customer talks to the camera, are the ones I find hardest to identify today. For me, this is the core of the problem. I’m not saying Smooche is cheating. I’m saying that even when I look closely, I can’t know, and neither can the average customer.
How to Use AI Without Losing Authenticity
I suggest a few rules.
- Show the real product and real results: A before and after should come from a real person, not from an artificial image.
- Say when it’s AI: There is no need for a label on every small detail, but a person who appears as a “witness” in a video should be real, or the video should say clearly that they are not.
- Ban urgency that isn’t real: A timer or a “last day” should match a real end to the promotion.
- Keep a human review: Someone on the team should read and check every AI-made text, image or video before it goes live, so mistakes or false claims never reach customers.
Conclusion: Innovation Needs Trust
AI can help a brand work faster and test more things. I’m not saying it is bad. But the better it gets, the more trust becomes the real value of a brand.
In my test, I couldn’t tell the difference between real and generated. Can you? And if a brand clearly said what was generated by AI, would it reassure you, or would it have the opposite effect?
About this article
This article has been written by a student on the Grenoble Ecole de Management’s Advanced Masters in Digital Strategy Management. As part of a content creation assignment, students are given the task of writing articles based on their digital interests and disseminating the articles online. Articles are marked but we make minimal changes to the content. Thanks for reading! James Barisic, Programme Director, MS DSM.
