AI deepfake videos are giving new relevance to the saying “if it looks too good to be true, it probably is.”
From a recent analysis of videos on TikTok and Instagram, tech experts from the cybersecurity company NordVPN report a surge in videos promoting beauty products with AI-generated endorsements and results.
"Deepfake scams used to focus on impersonating public figures to push crypto and investment schemes. Now, they are targeting everyday shoppers with fake endorsements for skincare or consumer goods," says Marijus Briedis, Chief Technology Officer at NordVPN.
"These videos are designed to feel familiar and trustworthy, which is exactly what makes them dangerous."
Worse, their creators are figuring out ways to circumvent labels on Instagram and TikTok marking them as AI-generated content, he says.
A deepfake is a piece of digitally manipulated audio or video that create false depictions of real life.
Many deepfakes on social media mimic podcasts, interviews, and genuine product reviews, resulting in content that feels like a real recommendation and not an advertisement, and designed to make viewers more likely to trust it.
Other methods of using AI to fake reality include stealing content from real people and swapping their faces out with an AI-generated version.
The products being promoted may not necessarily be harmful, the tech giant says, but the claims made about them are wildly exaggerated — with AI-generated voices reporting dramatic results, praising customer service, or urging viewers to buy before the product sells out.
These videos are often supported by fake or coordinated comments underneath, creating what Breidis calls “a loop of false trust.”
How to Spot Deepfakes on Social Media
"There are several warning signs to watch for,” Breidis says. “Sudden changes in accent or tone, unnatural facial expressions, and vague claims without credible sources should all raise suspicion."
Another caveat: creators are figuring out ways to circumvent AI tags on posts, he says.
"We observed a particularly convincing deepfake promoting Korean skincare on TikTok. It managed to avoid being labeled as AI-generated because a real person introduced the clip,” Breidis says.
The comments underneath the video appear to be a mix of real and fake consumers.
If a video creates urgency, promises unrealistic results, or discourages independent research, it should be treated with caution, Breidis warns.
Before making a purchase through TikTok Shop, review the seller's profile, ratings, and refund policy, he says. If the product does not meet expectations, TikTok's buyer protection allows customers to request a refund directly through the platform.
Some creators are already helping to raise awareness. In the example below, content creator John Dorsey breaks down how to tell a product recommendation video is AI-generated.
Steps to Take if You’ve Been Misled
- Report the content to the platform so it can be reviewed and removed.
- If you clicked through to an external website, shared payment details outside of a trusted platform, or never received your order, contact your bank or payment provider immediately and change passwords on any accounts that may have been compromised. (Paying with a credit card offers stronger financial protections than using a debit card.)
Keep records of the transaction, the video, and any communication, and report the incident to the FBI's Internet Crime Complaint Center.