Photos of children no longer belong on the internet.

The two photos were taken as typical teenage selfies: looking in the mirror, fully clothed. But when online predators got hold of the images and ran them through an artificial intelligence image-generating tool, they became the basis for extreme pornographic videos.

These examples come from Report Remove, a service that allows children whose explicit photos have been distributed without their consent to confidentially report the image and request that it be blocked or removed from social media. Due to breakthroughs in artificial intelligence and the widespread availability of AI models and apps for “removing” clothing from photos, some minors are becoming victims even without any contact with criminals.

These two cases, according to the Guardian, show why the UK's National Crime Agency (NCA) and the Internet Watch Foundation (IWF), an organization for protecting children online, have issued guidelines advising parents to consider limiting the visibility of family photos, for example by making social media profiles private and sharing photos of children only in "close friends" groups.

The guidelines also recommend reviewing social media accounts for old photos that predators could exploit, as well as reconsidering consents for the use of photos, for example with schools or sports clubs, which may have been signed before advances in artificial intelligence enabled image manipulation.

The guidelines outline three measures: checking privacy settings on social media accounts; reviewing who can see images of their children; and having open conversations about giving people and organizations permission to post images online.

“The average parent or carer doesn’t post a photo of their child online thinking it could be downloaded and turned into child sexual abuse material,” Lorna Sinclair, the NCA’s child sexual abuse education manager, told The Guardian. “There are a lot of parents and carers who don’t know this problem exists.”

The publication of the guidelines also followed cases in which school websites in the United Kingdom were targeted by blackmailers, who downloaded photographs of children, used AI tools to turn them into child sexual abuse material, and then threatened to publish the results.

Both organizations say they are not telling parents what to do, but rather that they want them to be aware of the problem and know what steps they can take. Data shows that artificial intelligence is a growing source of child sexual abuse material. The amount of AI-generated child sexual abuse material found online rose by 14 percent last year, according to the IWF, which identified 8.029 realistic images and videos of such material created using AI in 2025, the British newspaper reports.

The average parent or guardian doesn't post a photo of their child online thinking it could be downloaded and turned into child sexual abuse material. There are many parents and guardians who don't know this problem exists, Sinkler said.

Dan Sexton, chief technology officer at the IWF, told The Guardian that he feels "very uncomfortable" telling parents not to post photos of their children publicly, but feels he has no other option because, according to him, there is not enough protection from this technology.

United Nations Secretary-General Antonio Guterres warned on Monday that artificial intelligence is developing faster than anyone can keep up, calling for globally agreed rules to reduce potential risks - especially to children.

"Technology that can reshape economies, transform the world of work, influence elections and shift the balance of security is being introduced faster than anyone, including the people who create it, can keep up," Guterres told delegates at the first-ever intergovernmental global dialogue on artificial intelligence in Geneva.

“Innovation needs safeguards... If artificial intelligence is to be powerful, it must be regulated,” Guterres told delegates.

Guterres stressed that globally harmonized rules on artificial intelligence must prioritize the safety of children, following examples in which minors were directed towards self-harm and deceived by machines posing as friends.

“We don’t let a drug reach a child until it’s proven safe. We test every toy. Yet artificial intelligence has reached our children – their learning, their friendships, their most private questions – before anyone asked what it would do to them,” he said.

He called for a Child Safety Pledge in Artificial Intelligence, according to Reuters, under which companies developing systems would have to prove they are safe before making them available to children. The systems would also not be allowed to generate sexual images of children, and when a child shows signs of distress, the system would be stopped and connected to a human who can help.

The UK government is already introducing restrictions, including making it illegal to possess, make or distribute AI tools designed to generate child sexual abuse material. It is also giving tech companies and child protection agencies the authority to test whether AI tools can produce child abuse images.

Sexton acknowledges that the long-term solution may not be to remove photos from public social media accounts. He wants legislation for AI models that would require them to be “secure by design,” a principle invoked by internet safety activists who want social media platforms to be built responsibly, with the well-being of users as a primary goal. “People use the term ‘security by design,’ but I don’t see that here,” he told the Guardian.

A government spokesman said UK law is “clear” about the illegality of creating, possessing or distributing child sexual abuse material, including content generated by artificial intelligence, and that such material must be proactively removed from technology platforms under the Internet Safety Act.

The spokesman added: "We will not hesitate to go further to protect children."

Sexton says pedophiles use various AI tools in combination to create images and videos, including “open source” AI models that can be freely downloaded and customized by users.

The IWF says it has already become almost impossible to distinguish between real child sexual abuse images and AI-generated material. Such content is also illegal in the UK, but this means authorities may struggle to distinguish victims who are in real danger from deepfake content.

AI videos, Sexton said, can still be distinguished from reality with the help of IWF analysts, but generally the only way to determine the origin of the material with certainty is to see if pedophiles on the dark web are taking credit for certain images.

“The only way we know is when we see the authors pointing to what they created,” he says, adding that distinguishing AI content from reality could remain an ongoing struggle.

“It's a problem that needs to be solved. But I don't know if it can be solved at the level of accuracy that we need, because the technology is constantly changing,” he says.

Lorna Sinkler, the NCA's child sexual abuse education manager, said the agency's new guidelines aimed to highlight a danger that many parents and carers don't know exists. Perpetrators are repeating a familiar pattern - they are among the first to take advantage of technological changes.

“As a society, we are still learning about artificial intelligence and how it is used in everyday life,” she says. “We are learning about the benefits, but also how this technology can be misused. The reality is that perpetrators will always be among the first beneficiaries of technological advances.”