Deepfake Video Detection: 60 Faked, One Security Sign

A 74-year-old retiree in the Netherlands is about to stand trial for something that sounds almost too strange to be real: prosecutors say he made deepfake sex videos of roughly 60 well-known Dutch people, including a member of the royal family. Not one video. Not one victim. Sixty. And police say he did it alone, from wherever he was sitting, using tools that are free and take about 25 minutes per video. This is why deepfake video detection stopped being a "someday" problem and became a "today" problem, and it's why this case deserves way more attention than it's getting.

TL;DR: Deepfake video detection matters because one 74-year-old retiree in the Netherlands is being prosecuted for allegedly making fake sex videos of 60 well-known people, proving that a single person with no special skills can now industrialize this kind of harm using artificial intelligence.

Let's slow down on that number for a second: 60 people. According to NL Times, the victims include Princess Amalia, television presenter Hélène Hendriks, and former police spokesperson Ellie Lust. These aren't random strangers pulled from a stock photo site. These are real, recognizable people whose faces were taken from real photos, real video, real public life, and stitched onto sexual content they never agreed to appear in. If someone can do that to a princess, they can absolutely do it to your sister, your coworker, or you.

Here's the part that should genuinely worry you: this isn't a hacker with a server farm in a basement. It's a retiree. One guy. No team, no company, no special training that we know of. An AI expert quoted in the deepfake research around this case put it bluntly, the barrier to making this kind of content has basically disappeared, meaning anyone, regardless of age, education, or profession, can make these videos now. That's the whole story in one sentence. The "genius hacker" myth is dead. What's left is a much scarier reality, which is that scale without accountability is the actual deepfake threats facing ordinary people today.

What deepfake video detection means for real people facing deepfake threats, not just tech companies

When people hear "deepfake video detection," they picture some lab coat squinting at pixels. In practice, it means something much more personal: figuring out, quickly and confidently, whether the media someone sent you, meaning the video, image, or audio, is really of the person it claims to be. That question used to matter mostly for celebrities and politicians. Now it matters for regular people, because the artificial intelligence tools that make fake video have gotten cheap and the detection models built to catch fakes haven't caught up.

The Netherlands treats this as a real crime, not internet mischief. Creating and distributing sexual deepfakes without consent is illegal there, and the suspect in this case faces up to two years in prison. That matters because it draws a legal line: your face is not public property, and turning it into sexual content without your permission is abuse, full stop. According to Law & More, Dutch criminal law already has provisions built for exactly this kind of harm, treating it in the same family as revenge pornography. This article is part of a series, start with Age Verification Roblox 31 Lawsuits Test Section 230.

How deepfakes spread faster than the truth can catch up, and what datasets like the DFDC and UADFV dataset teach us

The uncomfortable math is this: making a fake video takes 25 minutes and costs nothing, but explaining that it's fake takes a lot longer, and by then it's already been screenshotted, downloaded, and reposted somewhere you'll never find it. Speed favors the liar every single time. Researchers building detection models often train them on public benchmarks like the datasets DFDC and the UADFV dataset, which collect thousands of real and fake clips so machine learning systems can learn the subtle patterns that separate a genuine clip from a manipulated one.

That number should stop you cold. This isn't mostly political misinformation or fake celebrity endorsements. It's overwhelmingly sexual, non-consensual content, and the people targeted are overwhelmingly women. The Dutch case fits that pattern exactly: a presenter, a spokesperson, a princess, all women, all turned into content without a say in the matter.

Prosecutors say the suspect created and distributed the videos over roughly a year, involving about 60 well-known Dutch figures, before investigators identified him.

reported by NL Times

Look, nobody's saying prosecution alone fixes this. Critics have a fair point: one conviction in the Netherlands doesn't stop someone in another country from doing the exact same thing tomorrow, using the exact same free tools. Content crosses borders in seconds; court cases take months or years. But the counterpoint matters too, prosecution establishes that there's a real cost to this behavior, and in countries with strong rule of law, that deterrent effect is not nothing. It also sends a signal to victims: this is worth reporting to the press and to police, because someone will actually listen.

---

Deepfake video detection tools versus what a trained eye and detection models can still catch

So can software just catch this stuff automatically? Not reliably, not yet, and anyone telling you otherwise is overselling. Here's a straight comparison of where things actually stand.

What people assume	What's actually true right now	Status
Deepfake video detection software catches every fake video automatically	No commercially available tools fully catch every fake; detection models are a moving target as fakes improve	Unresolved, 2026
Deepfakes always look obviously fake	Many fakes are convincing at first glance, though skin can appear too smooth in lower-quality generations and lighting can look slightly off	Ongoing research
Only celebrities and public figures get targeted	Anyone with public photos, meaning nearly everyone on social media, can be turned into a deepfake using machine learning	Confirmed, 2026
Reporting a deepfake does nothing	Dutch prosecutors are actively pursuing criminal charges, showing enforcement systems are starting to respond	Prosecuted, 2026

Notice that middle row. It's a small tell but a real one: in a lot of lower-effort fakes, someone's skin can appear too smooth, almost plastic, and shadows don't fall quite right across the face. It's not foolproof. Better generation tools are closing that gap fast. But it's still one of the few things an ordinary person can use to detect deepfakes with their own eyes before believing (or sharing) a video.

Why no commercially available detection models can promise a perfect answer

There's a phrase worth remembering here: no commercially available tools can currently guarantee catching every deepfake, and any product or service that claims 100% accuracy is lying to you. Detection and generation are locked in a constant back-and-forth, where each improvement in artificial intelligence used to detect deepfakes gets studied and countered by whoever's building the next generation tool. That's not an excuse to give up. It's a reason to build habits, not blind trust in any single service or system. Previously in this series: Biometrics Fake Passkey Texts Hijack Microsoft 365 Podcast.

Can deepfake video detection tell whether a photo or account is really who it claims to be?

Not fully, and not on its own, but it can get you most of the way there. Facial comparison and image-matching systems can tell you how closely two faces align, whether a profile photo shows up elsewhere online, and whether the media, meaning the video, image, or audio in question, has telltale signs of manipulation. It won't hand you a courtroom-ready verdict by itself. But it gives you evidence, and evidence is exactly what turns "I think this is fake" into something people actually believe.

If you've ever wondered whether a photo or profile is really who it claims to be, that's the exact question this kind of tech exists to answer. Here's one useful thing you can actually do, before you ever touch a paid tool or service: run a reverse image search on a suspicious photo or video thumbnail. It's free, it takes ten seconds, and it will often show you if the same media, or an earlier, unedited version of it, already exists somewhere else online. That single check has caught more fakes than most people realize, because a lot of deepfakes still start from a real photo that's floating around somewhere with its original context intact.

What audio deepfakes, voice cloning, and synthetic media add to the picture

It's not just video. Audio cloning is exploding right now too, and according to tovima.com, one Greek family was recently targeted by a scam using a cloned voice of a relative. Cloning software for audio is now sold for as little as $500, according to reporting from MLive.com, which means voice fakes are following the exact same cost-collapse curve that video deepfakes already went through. The systems behind both threats are converging: fake video, fake audio, fake identity documents, all forms of synthetic media built on the same underlying artificial intelligence, all getting cheaper by the month.

Government agencies and security researchers are watching this too, because the same technology used against a Dutch retiree's alleged victims can just as easily target a company's finance department or a country's national security apparatus. That's not fear-mongering. It's the plain math of a technology that got cheap before it got safe.

---

Here's the question I can't stop turning over: this Dutch case only exists because someone finally got caught. For every retiree who gets identified and prosecuted, how many more are sitting somewhere right now, 25 minutes and zero dollars away from making video number 61, 62, 63? The technology didn't create a monster. It just handed ordinary people the tools to become one, quietly, at scale, from a laptop. That's the part nobody wants to sit with, and it's the part that should keep you a little bit alert every time a video of someone you know shows up looking just slightly too convenient to be true. Up next: Age Verification Roblox 31 Lawsuits Test Section 230 Podcast.

deepfake video detection: Frequently Asked Questions