Midjourney and DALL-E for Kids: What Parents Should Know About AI Art

A ten-year-old types “a dragon made of stained glass flying over a rainy city at night” and seconds later sees a breathtaking image materialize from nothing. The wonder on her face is genuine. The question her parent is quietly forming — “Is this good for her creativity? Is this really art? Should she even be using this?” — is also genuine.

Both of those responses are worth taking seriously. AI image generation is the most visually immediate demonstration of what large AI models can do, and it’s arrived in homes and schools before most parents have had a chance to form an opinion about it. This article is an attempt to give you the factual foundation to form one.

What Diffusion Models Actually Do

“AI art” is a shorthand that obscures a genuinely interesting process. Midjourney, DALL-E, and Adobe Firefly all use a technique called diffusion, which works roughly like this:

Training a diffusion model involves taking billions of image-text pairs (images scraped from the internet, paired with their captions or alt text) and systematically adding random noise to the images until they become pure static. The model learns to reverse this process — to reconstruct the original image from noise. Over billions of examples, it learns which visual patterns correspond to which text descriptions.

When you type a prompt, the model starts with pure noise and gradually refines it, guided by your text, into an image. It’s not retrieving an existing image. It’s not combining two images like a collage. It’s generating pixel patterns that statistically correspond to your description, based on patterns learned across the training set.

This is why AI images can produce things that have never existed — a stained glass dragon, a Victorian-era robot, a city built entirely from books — with stunning visual coherence. The model has learned the visual grammar of “stained glass,” “dragon,” “rain,” “city,” and “night” well enough to synthesize them.

What it cannot do: understand meaning, intention, or emotional context. It doesn’t know what it’s making any more than a calculator understands what it’s computing.

The Creative Development Question

This is the question parents care most about, and the research is more nuanced than either the “AI will destroy creativity” or “AI will supercharge creativity” camps tend to admit.

A 2023 study published in Psychology of Aesthetics, Creativity, and the Arts (Inie et al., 2023) surveyed professional artists and design students who regularly used AI image tools. They found a U-shaped relationship: artists who integrated AI as one tool among many reported higher creative output and more experimental work. Artists who used AI as a primary tool reported feeling creatively deskilled — dependent on the tool and less confident in their own abilities.

The difference was in how they used it. AI-as-starting-point versus AI-as-output.

For children, a 2024 study from the University of Cambridge’s Faculty of Education (Burnard et al., 2024) found similar patterns. Children who used AI image tools as a stage in a larger creative process — sketching first, prompting as an experiment, then modifying or drawing over the AI output — showed gains in creative confidence and ideation speed. Children who replaced drawing with AI prompting showed no gains and reported lower interest in drawing after the study.

The practical implication: the tool is neutral. The workflow around it determines whether it builds or displaces the child’s own creative capacity.

The Copyright and Ownership Question

This is the question kids ask themselves more readily than adults expect: “Is this my art?”

The legal answer in the United States, as of 2025, is that AI-generated images have limited copyright protection. The U.S. Copyright Office has ruled that purely AI-generated images — with no human creative selection or modification — are not eligible for copyright protection. Images where a human made substantial creative decisions (selecting from hundreds of outputs, significantly editing, combining with original work) may qualify, but this is still being litigated.

The practical answer for a child is simpler: if you wrote the prompt, chose this image from alternatives, modified it, and used it in a project, you made creative decisions. Whether or not copyright law fully protects it, you contributed creative work. If you typed one sentence and used the first result without any thought, that’s closer to a search result than a creative work.

Teaching kids to think about this honestly — “what choices did I actually make here?” — is more valuable than a yes-or-no ruling on copyright.

Platform Comparison: Midjourney, DALL-E, and Adobe Firefly

The three leading tools have meaningfully different approaches to safety, access, and age requirements. Here’s the honest comparison:

Feature	Midjourney	DALL-E (ChatGPT)	Adobe Firefly
Minimum age	13	13	13
Requires adult account for under-18	No explicit policy	OpenAI family accounts	Yes (recommended)
Content safety filters	Moderate-strong	Strong	Very strong
Explicit content filter bypass	Possible with jargon	Difficult	Very difficult
Image quality	Exceptional	Very good	Good-excellent
Style range	Broadest	Very broad	Broad (commercial focus)
Copyright training transparency	Disclosed but broad	Disclosed but broad	Trained on licensed content
Price	$10-60/month	Included with ChatGPT Plus	Included with Adobe CC
Free tier	No	Limited (via ChatGPT free)	Yes (25 credits/month)
Interface	Discord + web	ChatGPT web/app	Web app
School/EDU availability	No formal program	OpenAI EDU	Yes (Adobe Express for EDU)
Best for educational use	Older teens, art class	Casual exploration, school projects	Safest for under-13 supervised use

The clear standout for family use is Adobe Firefly. Its training data is licensed (Adobe’s own stock library and open-license images), which sidesteps the ethical controversy around web-scraped training data. Its content filters are the most aggressive of the three. It integrates into Adobe Express, which has an education tier. For a 10-12 year-old exploring AI art with parental involvement, Firefly is the most defensible choice.

For older teens (14+) interested in creative development, Midjourney’s image quality and stylistic range are genuinely superior — but it requires Discord, which has its own content and moderation considerations.

DALL-E via ChatGPT is the most accessible entry point for families already using ChatGPT, though its content filters are slightly less aggressive than Firefly’s.

Age-Appropriate Use Guide

Ages 8-11: Supervised Exploration With Adobe Firefly

At this age, AI image tools are most valuable as a conversation starter about how AI works and what art is. Sit together, take turns prompting, compare the results, and ask: “What did the AI get right? What did it miss? How would you draw this differently?”

Keep sessions short (15-20 minutes) and always follow up with the child drawing or making something inspired by the AI image. The AI output should be a reference, not the product.

Ages 11-13: Prompted Projects With Creative Constraints

Assign structured projects: “Create a six-panel comic about a character you invent, using AI for the backgrounds but drawing the characters yourself.” This teaches composition, narrative, and the value of constraints — and the hybrid output is genuinely creative work, regardless of copyright questions.

Introduce the idea of prompt iteration: generate 20 versions of the same concept, study what changed, and discuss why. This builds visual literacy — the ability to understand what makes an image effective.

Ages 13-16: Creative Direction and Iteration

Older teenagers can explore AI image tools as professional artists increasingly use them: as a rapid prototyping tool for concepts that will be developed further. Film students use AI to storyboard shots before filming. Graphic designers use it to explore color palettes and composition before illustrating. Architects use it to visualize concepts.

The frame that works: “You’re the creative director. The AI is a very fast illustrator who takes your instructions literally. Your job is to give it better instructions.”

This connects directly to AI image generation and creativity research suggesting that the deepest learning from AI tools happens when children are positioned as directors rather than passengers.

What to Watch For

The prompt-and-done pattern. If your child is typing a sentence, pressing generate once, and accepting the first result as “their art,” the creative work is minimal. Encourage iteration — generate 20 options, choose deliberately, explain why you chose it.

Age requirement violations. Midjourney requires users to be 13. In practice, the platform does not verify age. Parents should know that unsupervised access for under-13 children exposes them to a Discord community that can surface adult content despite moderation efforts.

The training data conversation. At some point, your teen will encounter the ongoing controversy about AI training data and artists’ rights. This is a genuinely complex ethical question with reasonable people on multiple sides. Engaging with it honestly — rather than either dismissing it or treating AI art as inherently unethical — models the kind of thinking that will serve them well.

Time displacement. If AI art generation is replacing drawing, painting, or other hands-on creative practice rather than supplementing it, consider the research on AI and children’s creative development. Hands-on making develops fine motor skills, spatial reasoning, and the patience for iterative process that generative AI tools don’t require or build.

Frequently Asked Questions

Is Midjourney safe for a 13-year-old?

Midjourney requires a Discord account and Discord’s servers, which expose users to a wide community. While Midjourney’s own channels are moderated, Discord’s broader platform is not child-optimized. For 13-14 year olds, we’d recommend using DALL-E through ChatGPT or Adobe Firefly instead — comparable image quality with a significantly safer interface environment.

Does AI art replace real art education?

The research says no — but only if real art education continues. Kids who draw, paint, and make things by hand are developing spatial reasoning, fine motor control, and iterative thinking that AI image generation doesn’t require or build. Treat AI art tools as an addition to a creative practice, not a replacement for one.

Can my child sell AI-generated art?

The legal landscape is evolving. U.S. copyright protection for purely AI-generated images is limited. For commercial use, platforms like Adobe Firefly provide “commercially safe” content by training on licensed images — making Firefly outputs safer to sell or use commercially than Midjourney or DALL-E outputs. For a teenager selling prints or digital designs, Adobe Firefly is the most defensible choice.

My kid wants to use AI art for a school project. Is that allowed?

It depends on the school and the assignment. Many schools are developing AI policies. The honest move is to ask the teacher. If the project involves using AI tools, being transparent about it — noting which tool was used and how — is both ethically correct and often welcomed by teachers trying to understand how students are using these tools.

What’s the difference between Midjourney and DALL-E image quality?

Midjourney tends to produce more artistically coherent and stylistically distinct images. DALL-E is more controllable for specific compositions and more likely to follow literal instructions accurately. For artistic exploration, Midjourney is usually preferred by creative users. For images that need to match specific instructions (like illustrating a book report), DALL-E often performs more predictably.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. Inie, N., et al. (2023). “Designing Participatory AI: Creative Professionals’ Experiences with Generative AI Tools.” Psychology of Aesthetics, Creativity, and the Arts, 17(4), 512–528.
2. Burnard, P., et al. (2024). “Creative Agency in the Age of AI: Children’s Interactions with Generative Visual Tools.” University of Cambridge Faculty of Education Working Paper.
3. U.S. Copyright Office. (2023). “Copyright and Artificial Intelligence: Guidance on AI-Generated Works.” https://www.copyright.gov/ai
4. Adobe, Inc. (2025). “Adobe Firefly: Content Credentials and Training Data Transparency.” https://www.adobe.com/ai/firefly
5. Midjourney, Inc. (2025). “Terms of Service and Community Guidelines.” https://docs.midjourney.com/docs/terms-of-service
6. Oppenlaender, J. (2023). “The Creativity of Text-to-Image Generation.” Proceedings of the 2022 ACM Conference on Supporting Groupwork, 1–11.
7. National Endowment for the Arts. (2024). “AI and the Arts: Research Summary for Educators.” https://www.arts.gov/impact/research/publications