Artificial Intelligence and safeguarding: the key issues schools must understand now

• The Department for Education has expanded its guidance on artificial intelligence in schools, adding new modules and significantly strengthening the safeguarding material. Produced with the Chiltern Learning Trust and the Chartered College of Teaching, the updated resources now cover areas such as SEND provision, school operations, and a much broader treatment of safeguarding concerns.

• A major addition is the explicit recognition of cognitive offloading as a safeguarding issue. The guidance warns that when pupils rely on AI to think for them, it can reduce their willingness to seek help from real people. This isn’t only a teaching problem; it creates conditions where important disclosures may never reach a trusted adult. The update also highlights the risks of pupils forming emotional attachments to AI chatbots. Some tools are intentionally designed to encourage this, and for vulnerable children it can displace real relationships and undermine the human contact that effective safeguarding depends on.

• Existing warnings about deepfakes, AI‑generated child sexual abuse imagery, grooming through avatars or chatbots, and AI‑assisted extremist content remain in place and have been reinforced. The guidance makes clear that the government’s rules on incidents involving nudes and semi‑nudes apply equally to AI‑generated sexualised deepfakes.

• The legal landscape has also shifted. The Crime and Policing Act 2026 has now become law, introducing offences related to creating, adapting, supplying, or offering nudification tools—AI systems that digitally remove clothing from images of real people. This is the first time the tools themselves, not just the resulting images, have been criminalised. The offences are not yet enforceable, and even once they are, the tools will not simply disappear. Research from 2026 showed that Apple and Google Play had dozens of nudification apps between them, with hundreds of millions of downloads and significant revenue, and although platforms removed apps when alerted, new ones quickly emerged. The new law applies only to apps, leaving browser‑based nudification services untouched.

• Safeguarding risks remain very real. Smoothwall’s 2026 research found that more than a quarter of educators had encountered pupils using AI to create sexualised or abusive imagery, and nearly one in ten had dealt with a student producing a fake explicit image of a peer. Legal progress is welcome, but schools still face an active and ongoing risk that requires clear policies, staff training, and pupil education.

• The DfE is clear that child protection and online safety policies must now reflect AI‑specific risks, including deepfakes and AI‑generated imagery. Filtering and monitoring systems must be reviewed annually, and those reviews must explicitly consider AI use. Leaders are also advised to check whether any mental‑health‑related apps or online tools used with pupils are regulated by the Medicines and Healthcare Products Regulatory Agency—something many schools may not yet have considered.

• All updated materials, including four staff modules and a leadership toolkit, are freely available. Module 3, which focuses on safe use, is recommended for all staff. The leadership toolkit is intended to be worked through collaboratively as part of strategic planning rather than read once and set aside.

Please click here to view the full guidance here:

Source: Source: GOV.UK — Department for Education