You assume the threat of deepfakes is about adults, politicians, or celebrities. The real, immediate risk is in the school corridor and group chat, where AI-generated impersonations of children are becoming a standard tool for bullying. This is not speculative. The technology to create a convincing video of a child saying or doing anything is now trivial, fed by a constant stream of source material from family social media accounts and school events. The harm is profound, the defences are absent, and the legal system is looking the other way.
The source material is already public
Creating a deepfake requires source imagery and video. For a child, this library is already compiled and publicly available, often by their own family. Every birthday party video posted to a semi-public account, every school sports day clip shared in a parent group, every TikTok dance uploaded by the child themselves becomes training data. Bullies do not need sophisticated hacking skills; they need basic search ability and a willingness to scrape.
The output is a synthetic video that can show the victim saying vicious insults, confessing to embarrassing acts, or appearing to break social taboos. The believability hinges on the mundane reality of the source footage. A deepfake built from clips of a child in their own bedroom or garden carries a visceral authenticity that a fabricated studio shot cannot match. This turns the child's own digital footprint, a record of their life, into the weapon used against them. Parents often share this material with the best intentions, unaware they are building a potential attack surface. You should ask hard questions about any platform that handles your child's image, a process I detailed regarding school apps and student data.
Why school bullying policies fail
School anti-bullying policies are built for a different era. They are designed to address verbal taunts, physical altercations, or the sharing of embarrassing real photos. They are wholly unprepared for evidence that looks real but is entirely fabricated. The standard response protocol—investigate, verify, mediate—collapses when the "evidence" is the problem itself.
Picture a school administrator presented with a video of a student apparently making a racist remark. The policy demands they take the complaint seriously. Yet the accused student can only offer a denial against what appears to be conclusive video proof. The system lacks the forensic capability or the procedural framework to authenticate digital media. The bully understands this. The attack is not just the distribution of the false video; it is the weaponisation of the institution's own rules to punish the victim further. The resulting paralysis protects no one and often leads to both children being treated with suspicion, compounding the trauma for the genuine victim.
The nightmare of removal
When an adult discovers a deepfake of themselves, the path to removal, while difficult, has some established, if flawed, routes. For a minor, every step is harder. Most social media platforms require the reporting account to be held by someone over a certain age, typically 13 or older. A parent reporting on behalf of their young child may hit immediate automated barriers. Even after verifying guardianship, the process is slow.
Platforms' content moderation systems are tuned for broadly recognised policy violations like nudity or graphic violence. A deepfake of a child making a false confession may not trigger these automated flags. It requires a human reviewer to understand the context—that the person in the video is a minor, that the content is synthetic, and that its purpose is harassment. This is a labour-intensive judgement call platforms are structurally ill-equipped to make at scale. The content can spread across dozens of services and private chats faster than any takedown process can chase it. The general principles for initiating removal when you find a deepfake of yourself apply, but the practical and emotional burden on a family is magnitudes greater.
The legal gap is the point
The law provides the sharpest illustration of why this form of bullying is so effective. In many jurisdictions, legal recourse for deepfake abuse is tightly linked to either defamation (for adults) or child sexual abuse material. A non-sexual, AI-generated impersonation of a minor designed to bully them falls into a gap. It may not meet the high bar for criminal harassment. Civil defamation suits are costly, slow, and require proving specific damages, a daunting prospect for a family.
The bully, often a minor themselves, operates with de facto immunity. They know the school is confused and the police have no clear law to apply. This creates a perverse incentive. The most damaging deepfake from a child's social perspective—one that ruins friendships and social standing—is precisely the kind that carries the least legal risk for its creator. While the US has passed the Take It Down Act to help remove certain intimate imagery, its scope is specific. The broader problem of non-consensual synthetic impersonation, especially of minors, remains largely unaddressed by statute. Understanding the limited tools under laws like the Take It Down Act is essential for managing expectations.
A defensive posture for parents
You cannot make the risk zero, but you can make the attack harder and less effective. Your posture must be proactive, not reactive. First, audit and lock down the source material. Review the privacy settings on every platform where you or your child post videos or photos. Assume anything publicly viewable can be downloaded and used. Have explicit conversations with extended family about not sharing identifiable footage of your child without permission.
Second, educate your child on digital authenticity. Teach them that seeing a video of a friend saying something cruel is not proof. Explain what a deepfake is in age-appropriate terms. Frame it as a new kind of lie, one that uses computers. This gives them a cognitive defence when they encounter one. Third, know the reporting pathways before you need them. Familiarise yourself with the reporting centres for major platforms. Save the links. Understand what documentation you will need to prove guardianship.
Finally, build a supportive environment where your child knows they can come to you without fear of losing device privileges if they are targeted. The goal is to be the first person they tell, not the last. This principle of open communication is the bedrock of dealing with any AI-facilitated threat, as emphasised in guidance on handling AI sextortion attempts.
Questions people ask
How can I tell if a video of my child is a deepfake?
Look for inconsistencies the current technology still struggles with: unnatural eye blinking or a lack of it, hair that doesn't move quite right, a slight mismatch between the voice and the lip movements, or skin texture that seems too smooth. Listen to the audio for robotic cadences or strange pauses. However, the most reliable indicator is often contextual. Does the behaviour in the video contradict everything you know about your child? The content is likely the giveaway before the technology is.
What should I do first if my child is deepfake bullied?
Secure your child's emotional safety first. Assure them they are believed and not in trouble. Then, document everything. Take screenshots of the posts, usernames, groups, and any related messages. Do not engage with the bully or their parents directly yet. Use your documented evidence to report the content to every platform where it appears, following their specific reporting process for harassment or impersonation involving a minor. Inform the school, presenting the facts calmly and requesting their policy response.
Can the school expel a student for making a deepfake?
This depends entirely on the school's specific behaviour policy and whether it has been updated to address synthetic media. Many existing policies prohibit harassment and bullying but may not explicitly name AI-generated content. Expulsion would typically require the school to demonstrate a severe and persistent breach of policy. The challenge is often one of proof; the school must be confident in its attribution of the deepfake's creation, which can be technically difficult. You should review your school's published policies to understand their defined boundaries.
Are there laws being made to stop this?
Legislative movement is slow and fragmented. Some regions are amending existing harassment or privacy laws to include digital forgeries. Others are proposing new offences specifically for creating or distributing harmful deepfakes without consent, with enhanced penalties when the victim is a minor. However, lawmaking lags behind the technology. The most promising developments are not in criminal law but in creating statutory duties for platforms to remove such content upon verified report, especially for minors, and providing clearer civil right-of-action for families.
Close
The deepfake impersonation of a child is a perfect storm: low technical barriers, high emotional impact, institutional confusion, and legal silence. It weaponises childhood itself, using a child's own image to isolate and torment them. Treating this as a distant, adult problem is a mistake. The tools are in the hands of teenagers now.
Your defence is not a single tool but a posture. It combines prudent sharing, open conversation, and a clear-eyed understanding of the systems that are supposed to help. You must assume the material exists to be misused and that the institutions you rely on are not yet equipped to handle it. The goal is resilience—to ensure that when, not if, this new form of bullying appears in your child's world, it meets a prepared child, a knowledgeable parent, and a plan that focuses on their wellbeing over a futile chase for perfect justice.