,
8 tweets,
4 min read
Read on Twitter
Last month I my attention was drawn to something suspicious: A fake LinkedIn profile connecting to Washington think tank types and senior US officials.
Even weirder: Her face itself appeared to be fake.
apnews.com/bc2f19097a4c4f…
Even weirder: Her face itself appeared to be fake.
apnews.com/bc2f19097a4c4f…
For a while now people have been worrying about the threat of “deepfakes,” AI-generated personas that are indistinguishable, or almost indistinguishable, from real live humans. I think I may have caught an example of one in the wild:
apnews.com/bc2f19097a4c4f…
apnews.com/bc2f19097a4c4f…
Most of the concern I’ve seen over deepfakes has been articulated in terms of impersonation — i.e. can a politician be made to say something they never said. But even imaginary deepfakes are a problem, for example to generate untraceable avatars.
Katie Jones has vanished now — her profile was removed roughly 36 hours after I contacted LinkedIn about her. But here’s what it used to look like:
Experts like @HaoLi81 said it was really easy to identify “Katie” as a GAN when you examined the full image. But if you saw a request from this LinkedIn user, with a tiny photo thumbnail, would you refuse?
Any image forensics people want to play around with “Katie”? Here is the photo I pulled off of LinkedIn before her account vanished: dropbox.com/sh/l96cjoh0stk…
A few people have had a look, including Christopher Schmidt of thisrentaldoesnotexist.com
And then there’s @IntelMercenary who beat me to the punch — and is much, much better at identifying GAN features than I am.
