BMC Session 1: Regulating AI in assessment: evidence, accountability and transparency
The Future-Ready Assessment Ecosystem: Learners
Assessment should empower every learner to demonstrate what they know and can do. As technology continues to reshape education, organisations have new opportunities to create more personalised, accessible and engaging assessment experiences while maintaining fairness, validity and trust.
This week explores how assessment can better support learners through AI-supported personalisation, inclusive design, accessibility, and improved learner experiences. Join experts from across the assessment community as they examine practical approaches to designing assessments that meet the diverse needs of today’s learners while preparing for the challenges of tomorrow.
About this session
Panel discussion and audience questions
Chair: Graham Hudson
Panel:
- Maddy McKay, Ofqual
- Dean Seabrook, Qualifications Wales
- Matthew Wingfield, contributing an awarding-organisation perspective
As awarding organisations explore how AI could support assessment development, delivery, marking and decision-making, they must work within existing regulatory principles while responding to technology that is developing rapidly.
This session brings regulators and the awarding-organisation community together for a practical discussion about the evidence, accountability and transparency required when AI enters an assessment process. The panel will distinguish established requirements from emerging thinking and unresolved questions, while considering how organisations can innovate without undermining public confidence in qualifications and assessments.
The discussion will explore:
- How awarding organisations are beginning to use AI to innovate or improve existing assessment processes.
- How the sector can balance innovation with public confidence.
- Which existing regulatory principles are most relevant to AI-supported assessment.
- What evidence should exist before an organisation moves from exploration to operational or high-stakes use.
- How validity, fairness, accessibility and security should be evaluated.
- Where accountable human judgement must remain visible and what an effective audit trail should show.
- What constructive dialogue between awarding organisations and regulators should look like as new approaches develop.
- What might cause an AI use case to be stopped, revised or restricted.
Victoria Merrick will also be invited to bring observations from the earlier school leader and practitioner discussion into the audience conversation.
The session will focus on practical principles and open questions. It will not ask regulators to announce unpublished policy, comment on individual organisations or endorse particular products.

