By Kara Mcgrath

Recently, some corners of the internet have become convinced that tattoos offer another type of bodily autonomy: Protection from having your likeness replicated in videos created by generative AI programs. Being covered in ink, the claim is, gives you a defense against deepfakes, because the current iteration of AI-powered video programs can’t handle all that artistic detail.

According to the AI researchers I spoke to, it is kinda true that generative AI models have a tough time recreating tattoos in videos… at least, right at this moment. “There's a little bit of truth to it, but it's not quite right,” says Hany Farid, a professor of computer science at Dartmouth College in New Hampshire and researcher who specializes in digital forensics and deepfake detection and has at least one tattoo of his own.

For as good as generative AI has gotten, simple visuals are what it recreates best. “It’s not that it struggles with tattoos, per se,” Farid says. “It’s that it struggles with details.” In other words, a wardrobe full of Dan Flashes shirts would arguably keep you just as protected as a tattoo sleeve.

There are also fewer people with visible ink than without—only 22% of people in the U.S. have more than one tattoo. That means the model training data set of people with tattoos is much smaller than the set of people without them, says Saetbyeol Leeyouk, a critical technologist and graduate student at MIT Media Lab whose work focuses on the intersection of ethics and AI.

But both Farid and Leeyouk say there is a big caveat here: We are talking about the capabilities these programs have right now in August 2026. If you’re going to say “tattoos make it harder for generative AI to recreate your likeness,” you have to end that sentence with “today,” Farid says. “Tomorrow, next week, next month, it’s going to get better. We know this.” Leeyouk agrees, saying “Eventually, [the tattoo issue] will be solved.”