Opinion

What parents should demand from schools before signing contracts with ai tutoring providers

What parents should demand from schools before signing contracts with ai tutoring providers

I’ve watched schools rush into partnerships with AI tutoring companies the way parents once snapped up slots at popular after‑school clubs: eager, hopeful and sometimes a little short on questions. As a parent and a journalist who covers tech and education, I want to be frank: AI tutoring can be brilliant, but it also carries real risks. Before you sign anything, there are clear, practical demands you should put to your child’s school. These aren’t legal tricks or technophobic stances — they’re common‑sense protections to make sure the technology serves pupils, rather than the other way around.

Ask for evidence of educational impact

First question I ask is simple: does it actually work? Vendors will wave around outcomes and testimonials, but schools should demand robust, independent proof that the tool improves learning.

Here’s what to ask for:

  • Randomised controlled trials or independent evaluations, not just vendor case studies.
  • Clear learning objectives and benchmarks aligned with the curriculum (e.g. KS2 maths, GCSE English).
  • Data on differential impact — does this help disadvantaged pupils as much as their peers?
  • When a supplier points to research, ask who funded it. A study sponsored by the company can be useful but not definitive. I’ve seen examples where tools like adaptive platforms (Century Tech, for example) show promise in small pilots, but scale and classroom realities change results.

    Demand transparency about the AI and how it’s used

    “AI” is a marketing buzzword. Parents should insist on clarity about what that actually means in the product their child will use.

  • Which models power the tutoring? Is it a large language model from a major provider (e.g. OpenAI’s GPT family, Microsoft Azure OpenAI) or a proprietary system?
  • Is the model fine‑tuned with pupil data, and if so, how?
  • Are generated answers checked for accuracy? How often does the system hallucinate or give incorrect explanations?
  • Insist that the school require the vendor to publish a simple, parent‑facing explainer: what the AI does, where it gets information, and what happens if the system is wrong.

    Protect pupil data — insist on strict privacy and security clauses

    Personal data about children is especially sensitive. Under UK GDPR and data‑protection law, schools must be careful about third‑party processors. I always tell schools to treat pupil data like a family heirloom — handle it with contracts, encryption and tight access controls.

  • No sharing of raw pupil data with third parties for marketing or model training unless parents opt in explicitly.
  • Data minimisation: only the data strictly necessary for the tutoring service should be collected.
  • Encryption in transit and at rest, plus documented access control and logging.
  • Right to audit: allow the school (or an independent auditor) to inspect security practices and request breach notifications within 24 hours.
  • Ask for specific contract language: retention periods, deletion procedures when a pupil leaves, and confirmation that data won’t be transferred outside the UK/EU without adequate safeguards.

    Demand human oversight and clear teacher roles

    AI should augment teachers, not replace them. In classrooms where AI tutoring is introduced, teachers must be central to planning, moderation and assessment.

  • Teacher review: no final grades or progress records should be solely automated — a qualified teacher must review and sign off.
  • Boundaries: define what the AI can and cannot do (e.g. provide practice questions, not set formal assessments).
  • Training: the vendor should commit to staff training and ongoing support for teachers, not just a one‑hour demo.
  • When vendors promise “personalised learning at scale,” I ask how teachers will be freed to use the insights. If the technology ends up increasing teachers’ workload with false positives and admin, it fails kids.

    Safeguard fairness and guard against bias

    AI models can reflect and amplify bias in their training data. That can mean weaker support for pupils with certain accents, dialects, or learning needs.

  • Bias testing: demand evidence of testing across gender, ethnicity, socio‑economic status, and SEND (special educational needs and disabilities).
  • Accessibility: the product should meet accessibility standards (e.g. screen‑reader compatibility, adjustable text sizes, alternative formats).
  • Adjustments: ask how the system supports pupils with SEND or EAL (English as an additional language).
  • Insist that the contract includes a remediation plan if bias or unfair outcomes are detected.

    Protect against vendor lock‑in and preserve parental choice

    Contracts that tie a school to a single company for years, with proprietary formats, can be a problem. I advise parents to demand flexibility so alternative tools or manual approaches remain possible.

  • Interoperability: require that student data and progress reports be exportable in open formats (CSV, common APIs).
  • Opt‑out: parents should be able to remove their child from the system without penalty, and the school should provide an alternative learning plan.
  • Exit clauses: the school should have the right to terminate if the product fails to meet agreed KPIs, with clear data deletion procedures.
  • Ask for clear safeguarding and content moderation policies

    When pupils interact with AI — especially chatbots or generative tools — safeguarding matters. Schools should get explicit commitments on moderation, escalation and monitoring.

  • Logs and review: there should be logs of interactions that can be reviewed for safeguarding concerns, with retention limited to what’s necessary.
  • Escalation paths: clear processes when a child discloses risk or when the AI produces harmful content.
  • Age‑appropriate controls and filtering to avoid exposure to inappropriate material.
  • Negotiate transparency on pricing and hidden costs

    Some vendors offer low introductory fees and then push expensive add‑ons — reporting dashboards, extra support, or premium content. Ask for a full cost schedule and what is included.

  • Break down per‑pupil costs, implementation fees, training, and ongoing support.
  • Ask who owns the content created with the tool — does the school retain rights to exercises and student outputs?
  • Request regular reporting and parent communication

    Schools should commit to regular updates for parents: what the AI is being used for, what benefits are being seen, and how concerns are being handled.

  • Termly reports on usage, outcomes and any incidents.
  • Parent workshops or clear guides so families understand the tool and the opt‑out process.
  • DemandWhy it matters
    Independent evidence of impactAvoids hype; shows real learning gains
    Strict data protectionsKeeps sensitive pupil information safe
    Human oversightPrevents automation replacing teacher judgement
    Bias testing & accessibilityEnsures fairness for all pupils
    Opt‑out & exit rightsPrevents lock‑in and preserves choice

    When I speak to other parents and teachers, the same themes come up: we’re optimistic about tools that can boost practice and free teachers to do higher‑value work, but cautious about black‑box systems, untested claims and creeping data collection. If your school treats these demands as negotiable, it’s time to press harder — schools represent every child in the classroom, and those contracts should reflect that responsibility.

    You should also check the following news:

    Could targeted business rates relief reverse the collapse of independent cinemas in midlands towns?
    Business

    Could targeted business rates relief reverse the collapse of independent cinemas in midlands towns?

    I keep returning to the same image: a faded poster stuck to a window, the neon sign of a...

    How apple vision pro adoption by uk design studios could cut production budgets and where it falls short
    Tech

    How apple vision pro adoption by uk design studios could cut production budgets and where it falls short

    I’ve been watching colleagues in UK design studios experiment with the Apple Vision Pro since...