DfE guidance backs AI for drafting SEND support plans

TL;DR:

• Updated government-backed guidance says teachers, teaching assistants and SENCOs may use AI to draft initial support plans and staff guides for pupils with special educational needs and disabilities (SEND).
• The guidance, developed by the Chiltern Learning Trust and Chartered College of Teaching with the DfE, draws firm lines: staff must be “extremely cautious” with pupil data and must not let AI write statutory documents such as EHCPs or decide whether a child has a special educational need.
• First published in June 2025 and downloaded over 44,000 times, the refreshed modules add sections on operations, inclusion, administrative tasks and safeguarding.

Government-backed guidance has expanded its endorsement of AI in English schools, now explicitly covering the drafting of SEND support plans — one of the more sensitive administrative tasks in education. The update builds on 2025 advice about personalising learning and adds new modules, while repeatedly stressing that responsibility cannot be delegated to the technology.

Where the lines are drawn

The guidance permits AI for first drafts of support plans, template communications and staff guides, but warns that even seemingly anonymous information can identify a pupil when matched with other data. Decisions about whether a student has a special educational need, contributions to statutory education, health and care plans (EHCPs) without significant professional review, and any provision of social, emotional or mental health support are explicitly off-limits. Schools are also advised to help vulnerable children distinguish AI from humans, given the risk of emotional attachment to chatbots.

Beyond SEND, the modules cover translating materials for pupils with English as an additional language, drafting risk assessments and policies, summarising recorded meetings with participant consent, and budget analysis — each paired with a caution that final accountability stays with the school. A new safeguarding section warns of “cognitive offloading”, where pupils defer their thinking to AI, and flags that AI-detection tools produce false positives that can disproportionately affect EAL pupils.

Looking forward

The guidance reflects a now-familiar UK public-sector pattern: cautious permission rather than prohibition, with the burden of judgement pushed onto frontline staff. For school leaders, the practical challenge is operational — building the data-protection habits and review processes the guidance assumes, at a time when SEND caseloads are already stretched. Whether “extremely cautious” translates into consistent practice across thousands of schools, or becomes a liability when a draft plan mishandles a child’s data, is the question the guidance leaves to individual settings.