AI Image Generation Safety Guide
A parent's guide to AI image generation tools — including Midjourney, DALL-E, and Stable Diffusion — covering deepfake risks, CSAM generation, and body image concerns.
AI Image Tools is tools that use artificial intelligence to generate detailed images from written text prompts, used for art, design and social media content.
Official age
13+
We recommend
16+
Developer
Various (Midjourney, DALL-E, Stable Diffusion)
Risks
3
Overview
AI image generation tools allow users to create photorealistic images from text descriptions. The most widely known include Midjourney (accessed via Discord), DALL-E (integrated into ChatGPT), Adobe Firefly, and open-source tools such as Stable Diffusion. These tools have extraordinary creative potential but also pose serious risks when misused. They can be used to generate realistic non-consensual intimate images (deepfakes), content that sexualises minors, and manipulated images designed to humiliate or harass individuals.
How children use it
Children and teenagers use AI image tools for creative projects, fan art, memes, and experimentation. Many encounter them through ChatGPT's integrated DALL-E feature or through social media communities on Discord and Reddit. Some teenagers use them to create realistic images of peers without consent, including manipulated intimate images. Open-source tools available online have fewer safeguards than commercial products.
Main risks
Recommended privacy settings
Content Policy Adherence (Midjourney)
Location: Midjourney server settings → Community Guidelines
Set to: Standard (no explicit content enabled)
Midjourney's default settings block explicit content. Ensure your child understands that attempting to bypass these filters through creative prompting violates the platform's terms of service and may result in account bans.
DALL-E within ChatGPT — Safe Messaging
Location: ChatGPT Settings → Safety
Set to: Default safe settings maintained
DALL-E integrated into ChatGPT applies content filters by default. Review these settings and ensure your child understands why generating realistic images of real people — especially classmates — is unacceptable.
Avoid Open-Source Tools Without Safeguards
Location: Device parental controls / network filtering
Set to: Block access to uncensored Stable Diffusion interfaces
Unlike commercial tools, many open-source Stable Diffusion implementations have no content filters. Use network-level filtering or device controls to restrict access to these sites on children's devices.
Parent actions
Talk explicitly with your child about what AI image generation is capable of, including deepfakes and non-consensual intimate images, and why creating such content is both harmful and potentially illegal
Time: 20 minutes
Check your child's devices and apps for AI image generation tools and review any images they have generated
Time: 15 minutes
Use network-level filtering to block access to unmoderated open-source AI image generation websites on your home network
Time: 20 minutes
Related app guides
If you need to report this
In immediate danger: call 999. For non-emergency police matters, call 101.
Concerned about a child but it's not an emergency? NSPCC helpline 0808 800 5000. Childline for young people 0800 1111.
Myths & facts
Myth: These tools only make cartoonish or obviously fake pictures, so there's nothing to worry about.
Fact: Modern AI image generators can produce highly realistic pictures of real people, and free 'nudify' tools are built specifically to strip clothing from an uploaded photo of someone real — including a classmate. UK schools have already had to deal with pupils, some as young as ten, creating this kind of image of peers or even teachers.
Myth: My child would need to pay for a proper subscription to generate anything explicit.
Fact: Midjourney and DALL-E do apply content filters by default, but plenty of free, unmoderated image generators are just a search away and have no age check or filter at all. Cost isn't the barrier here — awareness and an upfront conversation are.
Myth: Making a fake image with AI isn't as serious as a real photo, so it's not that big a deal.
Fact: UK law treats it as seriously as it sounds. Since 6 February 2026, creating — or even asking someone else to create — a sexual deepfake of a real person without their consent is a specific criminal offence (Data (Use and Access) Act 2025), on top of the long-standing law that any sexual image of someone under 18, AI-generated or not, is illegal to make, hold or share.
Age by age
- 8-10
AI image tools aren't built with young children in mind and there's no reason for this age group to use them unsupervised. If they're curious about AI, a child-friendly drawing or creative app meets that curiosity without the risks.
- 11-13
Curiosity about AI art often starts here, usually via ChatGPT's image feature or a friend showing them Midjourney results on Discord. Keep it to reputable, filtered tools, stay nearby while they use it, and be explicit that generating a picture of a real classmate is never acceptable, even as a joke.
- 14-16
Teenagers can find uncensored, unfiltered tools with a simple search, and this is the age group most likely to have encountered a 'nudify' app or heard about one from friends. Have the direct conversation that creating a sexual image of a real person is a criminal offence in the UK regardless of who asked for it, and it's worth asking whether their school has addressed this with pupils.
- 17-plus
Older teens can use commercial AI image tools much as an adult would, but the same law applies at every age: creating or requesting a sexual deepfake of a real person without consent is a criminal offence. It's also worth a word about their own image rights — photos they post publicly can end up fed into someone else's AI-generated content too.
Frequently Asked Questions
Related Resources
Sources
Explore more
Get more app safety guides
Practical updates on apps and online safety. No spam.