They Only Need One Photo: How AI Is Fueling a New Wave of Child Exploitation

By Polly Tarbard, Every Child Online Safe & Smart Online Blogger.

In the last year, a disturbing use of AI has begun to emerge online. Preators are now using artificial intelligence to create fake sexual images and videos of children. These images, while entirely fabricated, look terrifyingly real. These AI deepfakes are being used to harass, blackmail and groom young people. The Internet Watch Foundation (IWF) has seen a 380% increase in confirmed AI-generated child sexual abuse material (CSAM) between 2023 and 2024. It is also important to note that of all the AI imagery seen, 98% showed the sexual abuse of girls.

This raises some serious questions. How are these images being created? Why are deepfakes so dangerous, and what can families and schools do to protect young people from this emerging form of online exploitation?

One of the most common forms of AI misuse is the creation of fake sexual images using photos of real children. Offender are using AI image generators and so-called 'nudifier' apps to digitally remove clothing or paste a child's face onto explicit material. AI cannot invent a new face from scratch; it always builds on real images or features it has been trained on. Therefore, predators will source innocent photos from social media or school websites and then convert them into sexually explicit images using AI platforms. These deepfakes can be extremely convincing; analysts warn that even trained investigators sometimes struggle to tell them apart from real abuse material. However, while these images may be synthetic, UK law is very clear: they are treated the same as real child sexual abuse imagery. It is illegal to create, share or even possess under the Online Safety Act 2023.

What are 'deepfakes'?

Deepfakes are synthetic media where a person's likeness in an image, video, or audio is convincingly replaced with someone else's. In practice, this often means swapping one face onto another body or manipulating a voice to mimic someone. The term originated from internet forums in 2017, when users would paste celebrity faces onto videos. The process of creating deepfakes an algorithm that is trained on many photos of a person's face (showing various angles and expressions) until it "learns" their features. This data is then used to generate new images or map that face onto someone else. For example, an algorithm can be fed hundreds of pictures of Girl A and footage of Person B, and then produce a video where Girl A's face appears convincingly on Person B's body. This technique can also produce still images (like fake "photos") or clone voices. Initially, such technology required specialised skills, but simple apps and online tools now make basic deepfake creation "scarily simple" for non-experts. Some tools can even "nudify" a photo; generating a fake nude image from a clothed picture. A well-made deepfake video or image can fool viewers into believing a person did something they did not. With children, this can be used to create non-consensual sexual content or false evidence for bullying and blackmail. The technology is evolving quickly. By mid-2024, the IWF noted the emergence of the first "astoundingly realistic" AI videos depicting child abuse, not just still images. In some cases, synthetic voices have even been used in scams to mimic parents or children, and could be used to groom young people through impersonation.

How are predators using these photos?

Online predators are already finding ways to exploit and weaponise this technology. They are using fake sexual images to blackmail children into silence, compliance, or further abuse. This tactic, known as "sextortion," often starts with an innocent selfie, taken from social media or sent in private. Offenders use AI tools to generate explicit fakes, then threaten to expose the child unless they send more (real) photos or videos. The emotional impact can be devastating. Children may feel trapped, ashamed, and too afraid to tell anyone, which is exactly what these abusers rely on.

In other cases, predators use AI to build false identities and manipulate children over time which is a process known as online grooming. With the help of AI chatbots, voice changers, and realistic profile photos, offenders can easily disguise themselves as teenagers. Some even use AI-generated avatars or voices to imitate real young people, making it appear as though a child is chatting with a peer rather than an adult. This deception allows groomers to gain a victim's trust, gradually drawing them into private conversations, sharing personal information, or exchanging photos.

Once a predator has established trust, they can use AI-generated images or fake profiles to normalise sexual conversation, introduce explicit material, or pressure the child into sharing their own photos. In some cases, these interactions escalate into live-streamed abuse or attempts to meet in person. Children often don't realise they are being groomed until it's too late, as the manipulation feels like friendship or romance. Because AI makes these personas more believable with consistent messaging, realistic photos, and even voice interaction, grooming is becoming harder to detect.

The real-world consequences of this kind of exploitation are profound. Even when the images are fake, the emotional damage is very real. Victims of sextortion or deepfake grooming often experience fear, shame, and isolation, believing they have done something wrong. Some develop anxiety, depression, or post-traumatic stress, while others withdraw from friends, school, or social media altogether. The NSPCC has warned that children (particularly girls) describe feeling "violated" and unsafe in their own skin after discovering fake sexual images of themselves online. In several UK cases, victims reported that the stress of blackmail and humiliation led to self-harm and suicidal thoughts. For families, the impact can be devastating: parents often feel helpless, confused, and frightened, struggling to understand how something so serious could happen with just a single photo.

While girls are most often targeted deepfake abuse and grooming can affect anyone. Children and teenagers who spend a lot of time online, share personal photos, or seek validation through social media are particularly vulnerable. This could also make teenagers who are LGBTQ+ more vulnerable, as they may use social media to find a community if they do not feel accepted in their own home. Children and teens who have a lower digital literacy or are socially isolated are also at risk, as these are factors that make them easier for predators to manipulate or coerce.

What can be done?

While parents and educators play a frontline role in protecting children, they are not fighting this battle alone. A number of UK organisations and government-backed initiatives are working to tackle the growing threat of AI-facilitated abuse. The National Crime Agency (NCA) has made online child sexual exploitation a top priority, with officers now being trained to recognise and investigate AI-generated material. The Internet Watch Foundation, which partners with both government and tech companies, is actively identifying and removing AI-generated child sexual abuse content from the internet. In 2025, the IWF uncovered a hidden section of a chatbot site hosting AI-generated child abuse scenarios. This led to major takedown efforts and renewed calls for stricter regulation of AI platforms. Meanwhile, CEOP, a division of the NCA, continues to offer a dedicated reporting service for children experiencing online grooming or sextortion.

While these organisations and Online Safety Act make the creation and possession of these images a crime, preventing them from being made should be our main concern. For schools, this means going beyond standard internet safety lessons. Pupils need to be taught how AI works, what deepfakes are, and how to spot when something online might be manipulated, ideally starting before they become active on social media. Safeguarding policies should now explicitly cover synthetic images and impersonation, ensuring staff know how to support victims if a fake image is created or circulated. Pastoral teams should also be trained to recognise signs of grooming, such as sudden withdrawal, distress after using a device, or fear around being "exposed." At home, families should create a space where children feel able to talk about anything they see online, including mistakes. This means avoiding punishment-based responses and instead focusing on reassurance, support, and practical steps forward. Checking privacy settings together, having regular check-ins about online activity, and teaching children not to share personal images (even with people they trust) can all reduce the risk. Most importantly, children should know they won't be blamed if something goes wrong, and that help is always available.