OUP Study: UK Teens Split on AI Homework Cheating
Key insights
- Only 44% of UK students aged 13-18 believe using AI to complete all homework counts as cheating, per Oxford University Press research from May 2026.
- Just 15% of surveyed students felt they had received adequate school guidance on appropriate AI use, despite 72% voluntarily opting out of AI during a written task.
- 77% of students want teachers using AI to support lessons, but only 20% want AI providing direct answers.
Why this matters
Student definitions of AI academic integrity are diverging faster than school policy can respond, and the 15% guidance figure shows institutions are not close to closing that gap. For edtech founders, the 72%/77%/20% data points define a clear product constraint: students accept AI as a support layer but reject it as an answer engine, which is a workable design brief for compliant tools. Oxford University Press moving to trial Plan Assist in this policy vacuum signals that publishers, not regulators or schools, may end up setting the default norms for AI in UK classrooms.
Summary
Only 44% of UK teenagers consider using AI to complete all homework cheating, but just 15% report adequate school guidance on where the line is. An Oxford University Press study surveyed 3,100 students aged 13-18 in May 2026, with a further 704 students across 20 schools contributing qualitative research.
Essentially: (Oxford University Press, UK schools) are far apart on defining acceptable AI use.
- 18% think even asking AI for tips constitutes cheating, revealing wide divergence in how students define academic integrity.
- 72% chose not to use generative AI when given the option during a written task, with many preferring to express their own views.
- 77% want teachers using AI to support lessons, but only 20% want AI providing direct answers.
Oxford University Press also announced a trial of Plan Assist, an AI tool for teachers to generate personalized lesson plans and classroom resources, moving to set norms before schools do.
Potential risks and opportunities
Risks
- Schools without clear AI policies before the next major exam season face academic misconduct cases with no consistent adjudication standard, exposing governors and exam bodies to legal and reputational risk.
- Oxford University Press's Plan Assist, deployed into a policy vacuum, could establish publisher-defined AI norms as a de facto standard, displacing any future government or regulatory framework.
- The 18% of students who view even asking AI for tips as cheating may face inconsistent enforcement where no formal policy exists, creating equity risks for students in stricter informal school environments.
Opportunities
- Edtech platforms offering scaffolded AI support rather than direct answers are positioned to win school contracts by aligning with the 77%/20% student preference split documented in the Oxford University Press study.
- Oxford University Press's Plan Assist trial opens a teacher-facing AI tools market that competing educational publishers have not yet occupied publicly, with first-mover advantage in lesson planning and resource generation.
- The documented 15% guidance gap gives AI policy consultancies and professional development providers a clear mandate to approach UK school networks before the next exam cycle forces the issue.
What we don't know yet
- Whether the 20 schools in the qualitative sample represent a cross-section of state and independent UK schools, or skew toward early AI adopters.
- Which specific generative AI tools students were offered during the voluntary written task, since tool choice could significantly affect the 72% opt-out rate.
- Whether UK exam authorities have issued or are planning enforceable AI use policies ahead of the next major examination cycle, given the guidance gap this study documents.
Originally reported by MadeForMums
Read the original article →Original headline: Oxford University Press Study: Only 44% of UK Teenagers Think Using AI to Complete All Homework Is Cheating — 15% Say Schools Gave Enough Guidance