PSA: I just paid $200 for a 'deepfake detection' tool that totally failed me

Last month, a video of my cousin saying some wild stuff started going around our family chat. I was sure it was fake, so I bought a $200 yearly sub to a tool called 'VeriScan' to prove it. It gave the video a 95% 'real' score, which made everyone believe it MORE. Turns out my cousin's friend made it as a dumb joke with a free app. Has anyone else been burned by a detection service that got it completely wrong?

4 comments

4 Comments

bettym1114d ago

Wait, didn't we all think these tools were getting good? I used to tell people to just run stuff through a detector. But after seeing that Wired article paulw53 mentioned, and now your story, it's clear they're just guessing. They can't keep up with the free apps. My friend had a fake audio clip of our boss, and two different services said it was real. It makes you wonder, what are we even paying for?

jamieb801d ago

Yeah, that Wired article @bettym11 mentioned really laid it out. These services are basically just looking for old signs of fakes, like weird blurring. If someone uses a new app or cleans it up a little, the detectors have no clue. You're paying for a guess that's often wrong.

wendy82014d agoTop Commenter

Ugh, that's so frustrating. @paulw53 is right, those tools are not reliable at all.

paulw5315d ago

That VeriScan score reminds me of a Wired article about detection tools failing.

-1