Make Deepfake Videos: Pa. Trooper Gets Prison for Face Swap

Make Deepfake Videos: Pa. Trooper Gets Prison for Face Swap

Full Episode Transcript

---

A Pennsylvania state trooper used his work computer, and state photo databases, to make deepfake pornography of real people.

---

Investigators say they found more than ten thousand

Investigators say they found more than ten thousand pornographic images on that machine.

A judge sentenced him to five to ten years in prison.

This is a hard story, so I'll say it plainly.

If your face is on a driver's license, this story touches you.

That photo sits in a government database you never had a choice about joining.

The man is Stephen Kamnik, a Pennsylvania State Police trooper.

According to the Philadelphia Inquirer, he was sentenced for possessing child sexual abuse material and for creating deepfake pornography with A.I. tools.

KYW Newsradio reports he reached the state's PennDOT driver's license and I.D. photo databases.

---

The real question is this

So the real question is this.

When the faces in fake images come from real records, how does anyone trace them back?

Start with where those faces came from.

A driver's license photo is meant to prove who you are.

Here, according to the reporting, it became raw material for fabricated sexual images.

Nobody in those photos agreed to that.

Most of them probably never knew.

That's the part that stays with me.

---

The article's commentary says this case is less

The article's commentary says this case is less about one bad actor and more about oversight breaking down.

He was a trusted officer, working on a state computer, with access to sensitive records.

For investigators, that means access logs matter as much as the images themselves.

For the rest of us, it's a reminder that a database is only as safe as the people allowed inside it.

The article cites research from Sensity A.I. on child abuse deepfakes.

It says more photorealistic deepfakes turned up in the first half of this year than in all of last year.

Children between seven and thirteen made up about four in five of the A.I.-generated abuse images identified.

Pause on that for a second.

---

It also changes how police find this material

It also changes how police find this material.

Investigators have long used something called hash matching.

That means giving each known illegal image a digital fingerprint, then scanning for copies.

But a freshly generated fake has no fingerprint on file.

So the old tool simply doesn't see it.

For parents, that means the safety net they assume exists has a hole in it.

The numbers reach into ordinary life.

The article says about one in seventeen teenagers has been targeted by a deepfake nude image.

---

That's likely a few students in every school

That's likely a few students in every school.

These cases land on investigators as harassment, child exploitation, or identity theft, sometimes all three.

There's a legal wrinkle too.

Courts haven't settled on one standard for accepting deepfake-detection results as evidence.

So a case can turn on proof that's hard to produce, and harder to defend.

Some experts say detection is the real bottleneck.

They argue the field needs deeper forensic signals and tracking of where content came from.

But the article makes a sharper point. Proving an image is fake answers one question. Investigators also need to ask whose face was used. That's a comparison problem, and it's about speed.

---