What Are Deepfakes? What Every Parent Needs to Know About the Growing AI Threat
A photo of your child does not need to be inappropriate to be misused. It could be a school photo. A picture from a birthday party. A sporting photo. A dance performance. A family holiday. A selfie. Or simply a photograph proudly shared by a parent or grandparent. Artificial intelligence can now take ordinary images and use them to create convincing fake photographs and videos of a person doing things they never did.
That includes creating sexually explicit images of children.
And this is not simply a theoretical risk. International research led by UNICEF's Office of Strategy and Evidence – Innocenti, ECPAT International and INTERPOL estimates that at least 1.2 million children across 11 countries had their images manipulated into sexually explicit deepfakes in just one year.
In some of the countries studied, the prevalence was approximately 1 in 25 children.
Think about that for a moment.
That can be the equivalent of one child in a typical classroom.
The photograph used to create the image does not need to be sexual. The child does not need to have taken or sent a nude. They may never have communicated with the person responsible. They may have done absolutely nothing wrong. AI has changed what can be done with an ordinary photograph of a child.
Parents need to understand what that means.
What are deepfakes?
A deepfake is an artificially created or manipulated image, video or audio recording that convincingly makes it appear that a real person did or said something they never actually did or said.
Using artificial intelligence, someone may be able to:
put a person's face onto another person's body
make someone appear to say words they never said
imitate someone's voice
make someone appear to be somewhere they have never been
generate fake nude or sexually explicit images
create fake videos involving a recognisable person
impersonate someone during a phone or video call.
The technology itself can have legitimate uses in entertainment, education and creative industries. The danger is what happens when the same technology is used without someone's consent. For children, this can include bullying, humiliation, sexual exploitation, grooming, impersonation, fraud and blackmail.
"But my child doesn't share inappropriate photos"
This is one of the most important misconceptions parents need to understand. A child does not have to take or share an intimate photograph for an intimate image of them to exist.
AI has fundamentally changed that equation. So-called "nudify" tools can take an ordinary photograph of a clothed person and generate a fake nude or sexualised version.
The original photograph might be completely innocent.
A school photo.
A sporting photograph.
A picture at the beach.
A dance photo.
A selfie.
A family photograph posted on social media.
That means the online safety question is no longer simply: "Has my child shared an inappropriate image?"
Parents also need to ask: "Who has access to images of my child, and what could those images potentially be used to create?"
That is a significantly different risk.
The statistics parents need to know
In 2025, researchers working on Disrupting Harm, a major international research project led by UNICEF's Office of Strategy and Evidence – Innocenti, ECPAT International and INTERPOL, surveyed approximately 11,000 internet-using children aged 12 to 17 across 11 countries.
The findings are confronting. UNICEF estimates that:
At least 1.2 million children across those countries had their images manipulated into sexually explicit deepfakes in the previous year.
In some of the countries studied, the prevalence was approximately:
1 in 25 children.
That can represent roughly one child in a typical classroom. Not one child somewhere on the internet. Potentially one child sitting among their classmates. One child whose photograph may have been taken and transformed into something sexual that never happened. One child who may then have to live with the knowledge that an explicit image appearing to depict them exists and could be circulating online.
The children themselves understand the threat.
In some countries studied, as many as two-thirds of children said they were worried that artificial intelligence could be used to create fake sexual images or videos of them.
These are not statistics about children voluntarily creating explicit material. They concern something fundamentally different.
Someone else can create the image.
A child can refuse to send a nude. They can avoid taking sexual photographs. They can follow the rules their parents have taught them. And someone may still be able to create a sexually explicit image that appears to depict them.
That changes the online safety conversation families need to have.
AI-generated sexual abuse of real children is already occurring at significant scale internationally.
It is no longer a hypothetical future risk.
Deepfakes sit within a much bigger online exploitation problem
The deepfake statistics become even more concerning when placed within the wider online environment children are navigating.
In further Disrupting Harm research covering 21 countries and approximately 21,000 children aged 12 to 17, almost 1 in 5 children reported experiencing at least one form of technology-facilitated sexual exploitation or abuse within a single year.
UNICEF estimated that across the countries studied:
more than 15 million children had been exposed to unwanted sexual content
around 9 million children had been asked to engage in sexual conversations or share sexual images against their will
around 4 million children had sexual images of themselves shared without their consent.
These figures relate to technology-facilitated sexual exploitation and abuse more broadly, not specifically to deepfakes.
But that distinction is important.
Deepfake technology is emerging within an online environment where millions of children are already experiencing sexual pressure, unwanted content, exploitation and non-consensual sharing.
AI gives people who want to exploit or humiliate children another tool.
How could an ordinary photograph become a deepfake?
Imagine a parent posts a photograph of their 10-year-old child after a sporting match. There is nothing inappropriate about the photograph. The child is smiling at the camera in their uniform. Someone obtains a copy of the image. AI technology could potentially be used to manipulate it, sexualise it, place the child's face onto another body or generate new synthetic imagery resembling that child. A manipulated image could then be circulated through messages, group chats, social media or websites.
The child never created an intimate image. They never consented to one. The event depicted never happened. But the humiliation, fear and impact on that child can still be very real.
That distinction matters when we talk to children: A fake image does not mean fake harm.
Deepfakes are not only about sexual images
Sexual deepfakes are among the most disturbing risks, but deepfake technology creates other threats families need to understand.
Bullying and humiliation
A child's face can be placed into embarrassing, degrading or offensive situations.
A fake video could make it appear that a child said something racist, cruel, sexual or threatening.
A fake audio recording could imitate their voice.
That content can then spread through school group chats or social media before the child has an opportunity to explain that it is fake.
And as AI-generated content becomes increasingly convincing, saying "I didn't do that" may not immediately undo the damage.
Sexual extortion and blackmail
Deepfakes can also be weaponised for blackmail.
An offender does not necessarily need a genuine intimate photograph if they can manufacture a convincing one.
A child might receive a message saying:
"Send me another photo or I'll send this to your friends."
"Do what I say or I'll post this."
"Pay me or everyone at school will see it."
For a frightened child, whether the image is genuine may become almost irrelevant. What matters in that moment is that it looks real, and they are terrified other people will believe it.
Deepfake scams and voice cloning
Deepfakes can imitate voices as well as faces. That creates another risk for the whole family. Imagine receiving a phone call. The person on the other end sounds like your child. They are crying. They say something has happened. They urgently need money. The emotional reaction is immediate because you recognise the voice. But the voice may have been artificially generated.
Similar scams can imitate parents, grandparents, friends, colleagues and other trusted people. Families therefore need to rethink an assumption many of us have relied upon our entire lives:
Recognising someone's voice is no longer necessarily proof that you are speaking to that person.
Tip: Create a family verification word
One practical response to protect your family from deepfake voice cloning is to establish a private family verification word or phrase.
Choose something memorable that only your family knows.
If someone calls claiming to be a family member and asks for urgent money, personal information or help, ask them for the verification word.
If something still feels wrong, hang up.
Then contact the family member directly using a phone number you already know.
Do not call a number supplied by the person making the suspicious contact.
This is a simple family safety measure that becomes increasingly valuable as voice cloning improves.
Deepfakes can also be used for grooming
Children are often taught: "Don't trust strangers online."
The problem is that strangers are becoming harder to recognise. AI can help people create convincing false identities using synthetic photographs, videos and voices. A profile that appears to belong to another 11-year-old child may not belong to a child at all. A photograph is not proof. A video is increasingly not proof. A voice is increasingly not proof. This makes one of the fundamental lessons of digital safety more important than ever:
People online are not always who they say they are.
Children should be cautious when someone they only know online asks for personal information, photographs, secrecy, private conversations or a move to another messaging platform.
Why children's photographs deserve greater protection
Parents have shared photographs of their children online for years. Deepfakes do not mean families need to panic or remove every photograph they have ever shared. But they do change the risk calculation. Once an image is accessible to someone else, complete control over it becomes difficult.
A person may potentially:
screenshot it
download it
copy it
repost it
edit it
use it with an AI tool
combine it with other information about your child.
Deleting the original post later does not necessarily remove copies that have already been made.
This does not mean every photograph of your child will be misused.
It means parents should make a conscious decision about who needs access to their child's image and information, rather than assuming an innocent photograph can only be used innocently.
How to protect your family from deepfakes
There is no setting that can completely eliminate deepfake risk.
But families can reduce unnecessary exposure and make children better prepared to respond.
Review who can see photographs of your children
Check your social media accounts.
Ask yourself:
Is my profile public?
Who follows me?
Do all of these people need access to photographs of my child?
Could I share this privately instead?
A private account reduces exposure, but remember that someone with legitimate access can still screenshot or redistribute content.
Review old posts
Look through older social media content. Ask:
Would I make this information public today?
If the answer is no, consider deleting the post or restricting who can see it.
Talk to relatives and friends
Your privacy settings cannot protect a photograph uploaded publicly by someone else.
Grandparents, relatives, friends, sporting clubs and community groups may all share photographs of children.
Set clear expectations.
For example: "Please check with us before posting photos of the children publicly."
Teach children that AI does not remove consent
Children need to understand that creating a fake nude of another child is not simply an AI joke.
Neither is forwarding one.
The fact that the image is fabricated does not remove the humiliation or violation experienced by the person depicted.
A useful family rule is:
If you would not have permission to create, photograph or share it in real life, AI does not give you permission to fake it.
Teach children not to forward harmful images
Children may receive manipulated images through group chats. Curiosity can make forwarding seem harmless:
"Look what someone made."
Teach them to stop the chain. Do not forward it. Do not repost it. Do not ask someone else to send it. Tell a trusted adult.
If sexual imagery may involve someone under 18, children should not be encouraged to download or redistribute copies as evidence.
Teach children to stop, check and verify
Children are growing up in an environment where seeing something is no longer sufficient evidence that it happened.
When something seems shocking, unusual or unbelievable, encourage them to ask:
Who posted this?
Where did it come from?
Could it have been changed or generated using AI?
Can I check whether it is real somewhere else?
Critical thinking is now an online safety skill.
What should you do if someone creates a deepfake of your child?
Your first response matters.
If your child comes to you because someone has created, shared or threatened them with a deepfake, avoid beginning with:
"Why did you post that photo?"
"I told you not to talk to strangers."
"Why didn't you tell me earlier?"
Your immediate priority is keeping your child talking to you.
Start with:
"You are not in trouble. We are going to deal with this together."
Then take action: Record useful information such as usernames, account names, URLs, messages, dates and platform details where appropriate.
However, do not download, save or redistribute sexual images involving children simply to preserve evidence.
Report the account and content through the relevant platform.
In Australia, intimate deepfakes may fall within image-based abuse protections and can be reported to the eSafety Commissioner.
Threats, blackmail, grooming and sexual exploitation may also need to be reported to police.
If a child is in immediate danger in Australia, call Triple Zero (000).
What if someone threatens your child with a fake image?
Treat the threat seriously even if you know the image is fake. Do not dismiss your child's fear by saying:
"Everyone will know it isn't real."
They may not. And your child may be terrified that their friends, classmates or teachers will believe the image. Instead, reinforce five things:
Do not pay.
Do not send another photograph.
Do not negotiate.
Do not agree to another demand.
Tell a trusted adult immediately.
The aim is to take the power away from the person making the threat.
What if your child created the deepfake?
Parents need to be prepared for this side of the conversation too. AI tools can make harmful behaviour feel detached from reality. A child may say:
"But it isn't actually them."
"It was only AI."
"It was a joke."
"Everyone was doing it."
The fact that an image is fake does not remove responsibility. If your child creates or shares a sexualised, degrading or humiliating deepfake of another child, treat it seriously. Talk about consent. Talk about the impact on the person depicted. Talk about what happens when an image spreads beyond the original group chat. And make one principle clear:
Another person's body, face and identity are not raw material for entertainment.
The lesson needs to be bigger than:
"Don't use that app again."
The lesson is: Technology does not remove your responsibility for how you treat another person.
Five deepfake safety rules every child should know
For children aged 8 to 12, the conversation does not need to involve complicated explanations about machine learning or generative AI.
Children need practical rules they can remember.
1. A photo, video or voice online might not be real.
2. Never use AI to make embarrassing, nude or sexual pictures of another person.
3. Never forward a harmful or embarrassing image of someone else.
4. If someone threatens you with a photo or video, real or fake, don't give them what they ask for. Tell a trusted adult.
5. You can always ask for help when something online makes you scared, worried or uncomfortable.
Where Australian families can get help
If your child is affected by a deepfake, image-based abuse, cyberbullying, sexual extortion or other serious online harm, the Australian eSafety Commissioner provides information and reporting pathways for families.
Intimate images that have been digitally altered or faked to look like a real person may fall within Australia's image-based abuse protections.
If threats, grooming, blackmail or sexual exploitation are involved, contact the appropriate authorities.
If there is an immediate threat to a child's safety, call Triple Zero (000).
Do not leave a child trying to solve the situation alone.
Sources and further reading
UNICEF, ECPAT International and INTERPOL, Disrupting Harm
The Disrupting Harm research programme examines technology-facilitated child sexual exploitation and abuse and includes nationally representative research involving children, parents and caregivers.
UNICEF, "Deepfake abuse is abuse"
UNICEF's findings on AI-generated sexual imagery involving children, including the estimate that at least 1.2 million children across the 11 countries studied had experienced their images being manipulated into sexually explicit deepfakes within a year.
UNICEF, AI and child sexual abuse and exploitation
UNICEF's guidance examines how generative AI is changing child sexual exploitation and abuse, including the creation of synthetic sexual imagery and the implications for prevention, education and child protection.
Australian eSafety Commissioner, Deepfakes
Australian guidance explaining deepfakes, their risks and the challenges associated with AI-generated content.
Australian eSafety Commissioner, Image-based abuse
Information and reporting pathways for people experiencing image-based abuse, including digitally altered or fake intimate imagery.
Australian eSafety Commissioner, Sexual extortion
Guidance for children, young people and families responding to online sexual extortion and threats.