AI image generators are brilliant for classroom displays, worksheets and newsletters — and one of the fastest ways for a school to breach data protection. The line is simple once you know where it is. Here is a safe-use guide for staff.

The one rule that matters

Never upload photos of pupils to an AI image tool. Not to make cartoon versions for a display, not to "tidy up" a class photo, not for a leaving gift. A photo of an identifiable child is personal data. Uploading it to a consumer AI tool means sending it to a third party outside your school's data protection arrangements, with no lawful basis and no consent that covers it. Most consumer tools' terms also allow the company to use uploads to improve their models.

This question comes up constantly — see AI-generated imagery risks in UK schools. The same applies to staff photos without consent.

What is safe

What is not safe

Got a question about AI images in your school?

Tell me what you're trying to sort out — a tool, a policy, a parent question, a workload problem — and I'll give you a straight answer.

No sales pitch, no newsletter. Just a straight answer from a school governor who does this for a living. Prefer to talk? Book a 30-minute chat. Need more than an answer? Tailored policies from £495, staff training from £395 and support from £175 a month — less than a supply day.

"But parents consented to photos"

Consent for a photo on a display or the school website is not consent to send that photo to an AI company. If a school genuinely wants to do this, it needs its DPO involved, a specific lawful basis, consent that names the use, an approved tool with proper data terms, and usually a DPIA. In most cases it is simpler and safer to generate an illustration instead.

Copyright and accuracy

AI images can reproduce recognisable artwork and characters, so avoid prompts that name artists, brands or characters for anything shared publicly. And check historical or scientific images for accuracy — AI often gets details wrong. See copyright and generative AI.

Safeguarding

Image generation is also a safeguarding topic: pupils can misuse it to create deepfakes or inappropriate images of peers. DSLs should cover it in online safety education and filtering reviews. See deepfakes: what DSLs need to know.

A simple staff rule for your AI policy

"Staff may use approved AI tools to generate images from text descriptions for school resources. Photographs of pupils, staff or other identifiable people must never be uploaded to AI tools."

Add it to your staff acceptable use agreement.

Where AskColin comes in

If staff are already uploading photos — and in most schools somebody is — you need the rule in writing and staff who have heard it. I include this in the staff acceptable use agreement in the KCSIE AI package (£495), and cover it in staff training (£395). Or just ask me about your specific situation — I answer the same day.

Frequently asked questions

Can teachers use AI image generators?

Yes, for generating images from text descriptions for displays, resources and newsletters, using approved tools. Photos of identifiable people must never be uploaded.

Is it OK to upload class photos to ChatGPT to make cartoon versions?

No. A photo of an identifiable child is personal data, and uploading it to a consumer AI tool has no lawful basis or covering consent and may allow the company to use it.

Which AI image tools are safe for schools?

Image features in Copilot or Gemini on school accounts, or design tools the school has approved, used only with text prompts.

Does photo consent from parents cover AI tools?

No. Consent for displays or the website does not cover sending photos to an AI company. That would need the DPO, a specific lawful basis, named consent and an approved tool.

Can pupils use AI image generators at school?

Only through tools the school has approved as age-appropriate, with supervision, and with deepfake and misuse risks covered in online safety education.

← Back to all articles Ask me a question →