Fake Friends

AI-generated faces are taking over LinkedIn

Not even the most professional of social networks is safe from spam accounts.


Thousands of LinkedIn profiles are using computer-generated faces for their profile pictures, NPR’s Shannon Bond reports. Researchers at the Stanford Internet Observatory found more than a thousand fake accounts with GAN-generated images in just a few weeks’ time.

The fake profiles weren’t all coming from one company. In fact, NPR found that more than 70 businesses appeared to have enlisted the help of fake profiles, though they may not have all done it knowingly. Researchers speculate that some may have hired outside marketing companies for help with sales that may have created AI-generated LinkedIn profiles without authorization.

The rise of fake faces — Smiling portraits fill social media, and it’s no secret that some of them are fake. Since the early days of generative adversarial networks (GAN), computer-generated faces have become very convincing and very, very easy to access. Want to see an image of a person that doesn’t exist? It’s as easy as navigating to thispersondoesnotexist.com. There are so many “this x does not exist”-type sites that an aggregator site called thisxdoesnotexist.com has been created to compile all sorts of AI-generated things, from ponies to food blogs to music videos.

There are some tell-tale signs of an AI-generated face (mismatched earrings, weird or blurry backgrounds, mangled body parts, unusual stray hairs) but most look pretty darn convincing, even to a trained eye.

AI-generated faces have spread lies for Russia and pushed anti-Western propaganda, and recent reports show that they also perform the comparatively innocuous (yet pervasive) task of pestering LinkedIn users for sales or promotional purposes.

How does LinkedIn filter its users? — After Stanford Internet Observatory researchers contacted LinkedIn about the fake phots, the company said it removed those which broke its rules — but according to NPR the platform “did not give details about how it conducted its investigation.”

The network’s policy forbids “false or misleading content,” “fake profiles,” and falsified information. “Do not use an image of someone else, or any other image that is not your likeness, for your profile photo,” its guidelines state. LinkedIn’s transparency report shows that it removed more than 15 million fake accounts in the first six months of 2021, most detected automatically.

Next time you receive a strange LinkedIn message trying to convince you to work somewhere, buy something, or click something, check for mismatched earrings or unusual, blob-filled backgrounds. They might not exist at all.