AI fake nude sextortion: the talk to have before it happens

A shield, standing for a family plan against online blackmail

For years the online safety talk with a teenager rested on one rule: never send a nude, to anyone, ever. It was good advice. It still is. But it quietly assumed that blackmail needed a real photo to work, and that a child who followed the rule was out of reach.

That assumption no longer holds. A school photo, a swimming gala picture on a public account, a profile image: any of these can now be fed into a cheap app that makes a convincing fake. The child never sent anything. The message arrives anyway, with a picture attached and a demand for money.

What the numbers say this year

The clearest regional picture comes from South Korea, which has counted this more carefully than most. In a 2025 police crackdown, deepfake offences were the single largest category of cyber sex crime, 1,553 cases or 35.2% of the total, according to figures presented at an Amnesty International Korea forum and reported by The Korea Times in August. Of the 3,557 people apprehended, 1,761 were teenagers. “With just a few prompts, offenders can generate videos to their specifications,” said Cho Seung-no, a team leader in the Seoul Metropolitan Police Agency’s cybercrime investigation division.

National Police Agency data reported by the Seoul Economic Daily on 11 October put deepfake and falsified video sex crimes at 1,737 in 2025, up 44.5% from 1,202 in 2024, with 805 more recorded in the first eight months of this year. The same report carries the number that should worry parents most: of 88,726 requests to remove victims’ material from January to August 2026, only 30,100 were actually taken down. About a third.

The United States, which publishes the biggest dataset, shows the same shape. The US Secret Service reported in August that the child exploitation tipline run by the National Center for Missing & Exploited Children (NCMEC) received an average of 137 financial sextortion reports a day in 2025, a 37% rise on 2024. Reports involving AI-generated child abuse material went from 4,700 in 2023 to more than 400,000 in 2025. One case the agency described began with a fake nude made using the victim’s face.

Singapore’s own data on this is thinner. The police advisory of 3 September counted at least 52 sextortion reports since August and losses of at least $134,000, but it was about adults approached on dating apps, and it did not mention AI or minors. The playbook it describes is the one aimed at children too: move the chat to another app, get or make an image, threaten to send it to family and friends, demand payment. The police noted that the extortion often continues after the victim pays.

[[photo:1]]

The gap that matters: “I’d tell my parents”

Here is the finding we think every parent should sit with. In a March 2025 survey of 1,200 young people aged 13 to 20 by Thorn, a US child-safety nonprofit, 1 in 17 said someone had made deepfake nudes of them. When asked, 62% said they would tell a parent if it happened. Among those it had actually happened to, only 34% did.

That drop from 62 to 34 is where the harm compounds. A child who stays silent is a child negotiating alone with a criminal, and paying does not end it. Our own view is that a fake makes silence more likely, not less, because the child’s first fear is that nobody will believe the picture is not real. Thorn also found that 16% of teens still see these images as “not real”, which is its own trap: the person sending the threat is counting on the image looking real enough to everyone else.

The protective move, then, is not more rules about what your child posts. It is making sure that, when the message comes, your child already knows exactly what you will do.

The five-minute talk, before anything happens

Pick an ordinary day, not one after a news story. Something like:

  1. Name it plainly. “There are apps that can make a fake naked picture of anyone from a normal photo. Some people use them to scare kids into paying money.”
  2. Say the sentence that closes the gap. “If anyone ever sends you a picture like that, real or fake, you will not be in trouble with me. Not for the photo, not for who you were chatting to, not for anything.” The Australian Federal Police’s guidance (January 2024) puts it as staying calm and reassuring them “that it’s not their fault”.
  3. Agree on what not to do. Do not pay, do not send anything else, do not reply to argue. “Sextortion is never a one-time payment,” as the Secret Service’s Joan Hoback put it.
  4. Agree on what to do. Screenshot the chat and the profile, do not delete anything, then come and find you. Both the Singapore police and the AFP give the same order: keep evidence, stop contact, report.
  5. Mention the warning signs. The AFP lists several worth repeating: a stranger who moves the chat from one app to another, a profile whose camera “isn’t working” so it cannot video call, and promises to delete content.

If it does happen

Stay calmer than you feel. Your first reaction is the thing your child will remember next time.

Then work through it. Report the account on the platform where it arrived and block it. Make a police report. In Singapore, our earlier guide on reporting AI-made images of your child covers the Online Safety Commission route for getting content taken down. For an image of someone under 18, NCMEC’s free Take It Down tool works anywhere in the world: it makes a digital fingerprint of the image on your own device, so the picture itself is not uploaded, and participating platforms use that fingerprint to find and remove copies. Its FAQ is written around photos that were taken, and it does not say whether it handles AI-made fakes, so treat it as one route alongside the platform and police reports, not instead of them.

Korea’s removal figures are a reminder that takedown is slow and incomplete, even for a government that is trying hard. That is exactly why the conversation comes first. Nobody can promise a child the picture will vanish. You can promise them they will not face it alone, and that promise only works if they heard it before they needed it.

For the classroom side of this, see our piece on deepfakes made by classmates.

Was this useful?

Sources

  1. Sextortion in the Age of AI · U.S. Secret Service
  2. As deepfake sex crimes surge, Korea's response struggles to keep pace · The Korea Times
  3. Police to Hack Offenders' Accounts to Delete Child Abuse Videos · Seoul Economic Daily
  4. Deepfake nudes are a harmful reality for youth: New research from Thorn · Thorn
  5. AFP advice for parents on how to protect children from online sextortion · Australian Federal Police
  6. Singapore police warn against scams involving intimate images and sexually explicit video calls · The Star
  7. Take It Down: Frequently Asked Questions · National Center for Missing & Exploited Children