Safeguarding, online safety & Prevent

Home › Safeguarding, online safety & Prevent

Your safeguarding duties apply to AI-assisted work exactly as they do to everything else. The policy makes the AI-specific risks explicit, and points every concern down the route staff already know, to your designated safeguarding lead.

What it is

The section that connects AI to your existing safeguarding framework: KCSIE’s four areas of online risk, the Prevent duty, and the Online Safety Act. It names the AI-specific behaviours that are prohibited and treats them as conduct matters, and, where relevant, as safeguarding matters.

Fabricated media and deepfakes. Creating or sharing fabricated media intended to mislead, harass, distress or harm anyone in the community is prohibited. Where such material is sexual it may be criminal, and AI-generated sexual imagery of a child is a criminal matter in every circumstance. It’s reported to the designated safeguarding lead, handled under the child protection policy, and referred to the police where required, in line with the Online Safety Act 2023 and the Ofcom Protection of Children code. Multi-agency action follows Working Together to Safeguard Children 2026.

Nudification tools. Making, adapting, supplying or offering nudification tools, which are designed to remove clothing from images of real people, is a criminal offence under the Crime and Policing Act 2026. Any encounter with such a tool involving a member of the community is a safeguarding matter, reported by the usual route.

Voice cloning and fraud. Staff must not use AI to clone or simulate the voice or likeness of any pupil, parent or colleague. And because cloned voices and fabricated messages are increasingly used in fraud, any unusual or urgent request, particularly one involving payments, credentials or personal data, is verified through a known channel before anyone acts on it, in line with the DfE cyber security standards.

Published images of pupils. Automated tools can harvest photographs of children from public webpages and social media, including your own, for use in fabricated or abusive material. The policy expects you to keep your publication of identifiable pupil images, and the consent behind it, under review with that risk in mind.

Why it’s important

The tools to fabricate convincing images, audio and video are now widely available and simple to use. The schools that struggle most when this lands are the ones that never named the risk in their policy. Making the reporting route explicit, and the same one staff already use, means an AI-enabled incident is dealt with as quickly, and as seriously, as any other.