(WTNH) — On the dark side of the World Wide Web, hackers are manipulating photos and videos to make people say and do things they haven’t done. They are called deep fakes.
Dr. Vahid Behzadan is a professor at the University of New Haven. He specializes in artificial intelligence.
“Right now it has become so easy that a single profile picture of you, can be enough to create a sufficiently precise and realistic deep fake,” Dr. Vahid Behzadan, University of New Haven.
A mother in Pennsylvania arrested for creating a deep fake of a rival cheerleader vaping on her daughters squad, to get her kicked off the team.
“I went into the car and started crying, that is not me on video, because I thought if I said it, that no one would believe me.”
There are other deep fakes of politicians, saying things they didn’t. A video of Tom Cruise, is not Tom Cruise, it is not a look-alike, but artificially manipulated photos and sound.
Dr. Behzadan adds, “So what we have here is a still image of you Bob, and there is a template for movement and behavior, which is a recording of President Trump, and I have superimposed you over the template and created this animation of you speaking. “
At the University of New Haven, they have an entire lab dedicated to just finding solutions for detecting artificial intelligence. And they say the biggest problem is that it is a cat and mouse game. As somebody comes up with a way to figure out it is a deep fake, the deep fakes just keep getting better.
“The research community is working really hard to come up with automated solutions to detect deep fakes, there are already some prototypes out there, being used by companies like Facebook or YouTube, but they are not 100 percent accurate.”
While experts say it’s nearly impossible to protect yourself from becoming a victim of a deep fake, there are ways to tell if the image has been manipulated. Zoom in and look at the ears and the back of the head for deformities, but even now, New computer code can take those out.
“One of the most recent indicators is that looking at the reflection in the eyes of the character, this is how difficult it’s becoming to detect the deep fake.”
So is making a deep fake illegal? Scott Driscoll, a former computer crimes detective, has started internet safety concepts. He says it depends on what you do with it.
“There was one report where a company got information from one of their executives saying send us this amount of money send us the money, when in fact it was all artificial intelligence and they lost a lot of money,” Driscoll says.
The hackers can target corporations or family members.
“If you receive a video from somebody that looks like a relative, ‘I need money, send it to me,’ ‘I’m in trouble’, you also have to start validating any information that you get.”