Planning and resource adaptation
Differentiating existing schemes of work for reading age, SEND need and EAL - the task that consumes evenings and is genuinely well suited to assistance with a teacher checking the output.
Industries
AI in education, without the safeguarding gamble
Teaching staff are already using AI. The question for a leadership team is no longer whether it happens, but whether it happens with a policy, a standard and a workload benefit you can actually point to.
The pressure right now
What we hear from education
- Staff using consumer AI tools with pupil information in the prompt
- Academic integrity policies written before generative AI and never updated
- Workload initiatives that add a tool without removing a task
- SEND and safeguarding paperwork growing faster than the admin capacity to handle it
- Governors and inspectors asking what your AI position is, in writing
Where it pays off
Six places AI earns its keep here
Not everything on this list will apply to you. Most organisations start with one and extend once it has been measured.
Feedback and marking support
Structured formative feedback drafted against your own rubric, reviewed and moderated by the teacher. Summative grading stays human, and the policy should say so explicitly.
SEND and pastoral paperwork
Drafting support for provision plans, referrals and reports from existing evidence in your MIS, with the professional judgement and the signature staying where they belong.
Parental and internal communications
First drafts of letters, newsletters and case correspondence in the school's voice, with tone and reading-age controls for your community.
Admissions and enquiries
Handling the volume of repetitive prospectus, open-day and application questions with grounded answers, escalating anything sensitive to a person immediately.
Curriculum and quality review
Synthesising work scrutiny, learning walk notes and assessment data into a readable picture for leadership, with the underlying evidence linked.
Where we would start
The first three moves
1
Establish what is already in use
A short audit across staff, including the tools bought departmentally, so the policy is written about reality rather than assumption.
2
Write the two policies properly
Staff acceptable use and student academic integrity are different documents with different purposes. Both need to be readable in five minutes.
3
Prove it on workload
One workload-heavy process redesigned and measured across a term, so the next decision is evidenced rather than argued.
Risk and regulation
The part most suppliers skip
Where the risk sits
- Pupil personal data entered into tools with unread terms
- Safeguarding disclosures surfacing inside an AI conversation with no escalation route
- Age-appropriateness and content boundaries for any student-facing tool
- Academic integrity: detection is unreliable, so assessment design has to carry the weight
- Accessibility duties under the Equality Act for any tool put in front of learners
How we handle it
Student-facing and staff-facing uses are governed separately, because the risks are not the same and a single policy blurs both.
Every deployment is designed against your existing safeguarding and information governance, including a named escalation route for disclosures - not a bolt-on afterwards.
We work to current DfE guidance on generative AI in education and to your trust or institution's own data protection position, and we write the evidence so governors can see it.
Where this fits
Where to go next
Questions
Questions from education
Not summatively, and we would advise against any supplier who says otherwise. Formative feedback drafted against your rubric and moderated by a teacher is a genuine workload gain; unsupervised grading is a fairness and appeals problem waiting to happen.
Detection tools are unreliable enough that we do not recommend basing sanctions on them. The durable answers are assessment design, in-class verification and an integrity policy that tells students clearly what is and is not permitted.
Yes, and it usually should be. One policy set, one approved toolset and one training programme across the trust, with room for phase-specific differences - it is far cheaper than each school solving it alone.
Start with an audit of what you already run
Two to four weeks to an evidenced picture of your AI use, spend and risk - and a ranked list of what to do first.