Policy paper

Ofqual’s approach to regulating the use of artificial intelligence in the qualifications sector

Updated 16 July 2026

Applies to England

The Office of Qualifications and Examinations Regulation (Ofqual) is the independent regulator of qualifications and assessments in England. We regulate on behalf of students of all ages and apprentices to ensure that qualifications are high quality, valid and fair.

The rapid development and increasing accessibility of artificial intelligence (AI) presents opportunities and risks for the qualifications sector.

AI has the potential to support the design, development and delivery of assessments and, where used responsibly, to improve efficiency and productivity, to support innovation, and to promote growth across the sector. However, its use must be carefully managed to ensure that standards are maintained and that qualifications continue to be trusted.

This publication sets out how Ofqual is delivering its approach to regulating AI. Specifically, we set out how we will:

• enable safe AI‑driven innovation and growth in the sector
• ensure our regulatory framework remains proportionate, transparent and effective
• continue to manage risks to qualifications and assessments

How we regulate the use of AI

Ofqual’s regulatory approach to AI has been in place since April 2024. It is grounded in the high-stakes nature of qualifications and the significant impact that results have on students’ progression and life chances, as well as on employers, higher education providers and other users of qualifications.

Our approach is underpinned by 5 objectives:

• ensure fairness for students
• maintain the validity of qualifications
• protect the security of assessments
• maintain public confidence
• enable innovation

These objectives remain central to how we consider the opportunities and risks arising from the use of AI in the sector.

To ensure AI is used safely and appropriately by awarding organisations, Ofqual takes a cautious, evidence‑led approach, recognising both the need to manage risks and to enable responsible innovation.

This reflects the high-stakes nature of qualifications and the importance of maintaining trust in assessment outcomes. The use of AI in this context is still emerging, and its implications for assessment are not yet fully understood.

We apply our existing, outcomes‑based regulatory framework to the use of AI, rather than introducing new AI‑specific rules. We expect awarding organisations to be able to demonstrate that any use of AI delivers valid, reliable and fair outcomes, and to manage the risks associated with its use effectively. This includes ensuring appropriate expert human involvement and oversight, quality assurance and accountability for outcomes. We have also made clear that certain uses do not meet our requirements and should not be used.

We will continue to engage closely with awarding organisations and other stakeholders to build our evidence base, provide clarity where needed, and ensure our approach reflects emerging opportunities and risks.

Supporting innovation and growth in the sector

We support innovation and growth in the sector through a combination of guidance, engagement and targeted regulatory activity. The following sections set out how this is being delivered in practice across key areas of AI use.

Understanding AI use and innovation in the sector

Ofqual’s approach to AI is informed by ongoing engagement with awarding organisations, other regulators, government departments and international experts. This engagement has helped us to build our evidence base, understand emerging practice, and identify opportunities and risks associated with the use of AI in the qualifications sector.

The core of Ofqual’s engagement has been a series of workshops with awarding organisations run throughout 2025 and 2026, with the aim of better understanding their experiences of developing, testing and deploying AI, including whether there are any regulatory barriers to its adoption.

These workshops demonstrated a cautious but strong and growing interest in the potential use of AI in qualifications and assessment. A number of priority use cases were identified, including:

• item and assessment generation (for example, creating exam questions)
• stimulus material generation (for example, creating images, video, and audio)
• mapping and classification (for example, mapping qualifications against frameworks)
• use of AI to support marking
• use of AI to support invigilation

Awarding organisations identified a number of key risks and constraints to use of AI across these use cases, including:

• accuracy and reliability concerns
• data security and confidentiality risks
• capability, including technical capacity and staff expertise

While specific unnecessary or unintentional regulatory barriers have not been widely identified, awarding organisations are seeking greater clarity and confidence as they develop and test new approaches.

These insights highlight both the potential for AI to support innovation and growth in the sector, and the importance of clear and proportionate regulation to ensure its safe and responsible adoption.

In response, we are focusing our work on both clarifying our expectations and supporting appropriate uses of AI across a number of key areas of qualification design, development and delivery. The following sections set out our approach to these areas, including where we are building our evidence base, providing guidance and developing our regulatory approach.

Use of AI in marking

The use of AI in marking is one of the most prominent areas of interest for awarding organisations. Through our engagement with the sector, including our workshops, we have seen growing exploration of how AI could support aspects of marking and quality assurance.

However, the use of AI in marking high-stakes assessments raises important questions about validity, reliability and fairness. It also raises wider ethical questions about the role of human judgement in consequential decisions about students’ futures, and about the trust students, teachers and qualification users place in the marking process. Marking decisions must be accurate and consistent, so far as possible, but they must also be accountable, capable of being explained and, where necessary, challenged.

To support our understanding in this area, in January 2026 we published a detailed paper on the principles of AI use in marking high-stakes assessments. This work was informed by engagement with assessment experts, including awarding organisations, and technical specialists. It has been well received in the UK and internationally and has been published as a working paper to support ongoing discussion as the technology and approaches develop.

We have made clear to awarding organisations that AI may not be used as a sole marker in any assessments that are part of regulated qualifications. This research is directly informing our policy development and will support us in providing further clarity to awarding organisations on the appropriate use of AI in marking. For further information on the importance of technical capability, fairness and transparency in this context, see the blog post Using AI in marking: why technical capability, fairness and transparency all matter.

Alongside this, we recognise the potential for AI to support marking of lower‑stakes and classroom-based assessment.

We have contributed to work led by the Department for Education to support the development of AI tools to assist teachers in marking classroom assessments. This includes providing a bank of expertly marked student scripts, with appropriate permissions, for the content store pilot, supporting work to improve AI marking and feedback tools. We will continue to explore opportunities to support the responsible use of data in this way, where it can improve efficiency while maintaining appropriate safeguards.

AI in assessment design (including item and stimulus generation)

The use of AI in assessment design, including item and stimulus generation, is a key area of interest for awarding organisations. Through our engagement with the sector, we have seen significant exploration of how AI could support more efficient development of assessment materials and enable new approaches to test design.

These use cases may support innovation in the design of assessments. At the same time, they raise important considerations around the validity and quality of assessment materials, including risks relating to accuracy, bias and the appropriateness of AI-generated content.

We have engaged extensively with awarding organisations to understand how AI is being used in this area and to identify emerging practice. Alongside this, we are undertaking research to build the evidence base on the use of AI in assessment design. This work will inform our policy development and support us in providing further clarity and guidance to awarding organisations on good practice in this area.

AI in assessment delivery (including invigilation)

The use of AI in the delivery of assessments, including in areas such as remote invigilation, is an emerging area of interest for awarding organisations. Through our engagement with the sector, we are beginning to see exploration of how AI could support aspects of assessment delivery and oversight.

These use cases may offer opportunities to improve efficiency and flexibility in how assessments are delivered. However, they also raise important considerations, including the potential for bias, impacts on fairness for different groups of students, and the need to ensure that assessment conditions remain appropriate and secure, including in preventing malpractice and preserving the integrity of assessments.

We are engaging with awarding organisations and other stakeholders to understand how AI is being applied in this context and to explore the associated risks and benefits. This includes considering how existing regulatory requirements apply, particularly in relation to fairness, accessibility and the integrity of assessments. We recognise in our regulatory framework the high bar required to ensure that all remote assessment evidence can be trusted as attributable to each learner. We have provided guidance highlighting our concerns about whether the sole use of AI in such circumstances could meet that bar, reflecting the need for appropriate human involvement and safeguards.

We will continue to develop our approach to the use of AI in qualifications, drawing on our ongoing engagement, research and evidence gathering.

Across all these areas, we will focus on:

• building the evidence base through research and engagement
• using this to inform policy development and provide clear guidance
• supporting safe innovation, including through engagement and the responsible use of data

Use of AI by students

The widespread accessibility of generative AI brings growing risks to some established forms of assessment. Supervised exams remain largely protected, as students are not permitted to access AI tools during exams, and practical assessments are similarly less vulnerable where outputs cannot readily be produced by AI systems. In contrast, non‑exam assessment, primarily coursework completed outside controlled conditions or involving extended written or portfolio‑based tasks, faces greater pressure.

Where non‑exam assessment exists, it is typically intended to contribute to students’ learning as well as being a method of assessment. The use of AI in coursework undermines that intended learning experience, and where such use is undisclosed, damages the integrity of the qualification, as marks may no longer fully reflect the student’s own work. Ofqual therefore takes the risk of AI‑related malpractice extremely seriously.

Ofqual gathers and publishes data on AI misuse specifically. While the number of detected cases remains relatively low, there is significant concern among teachers, school and college leaders and others about the potential scale of misuse. This has informed our immediate use of regulatory tools and our consideration of longer-term changes to assessment.

Ofqual has robust rules and guidance in place requiring awarding organisations to prevent, investigate and manage malpractice. Awarding organisations are responsible for managing the risks associated with AI misuse and for assuring the authenticity of student work.

Ofqual has taken a number of steps to mitigate risks to qualifications and assessments. These include:

• collecting data on AI-related misuse to understand emerging risks
• publishing an advice note on AI-related malpractice and assessment to support awarding organisations in identifying and managing risks
• supporting the development and regular updating of the Joint Council for Qualifications (JCQ) guidance to help centres ensure the authenticity of student work
• actively communicating that students must not present AI‑generated work as their own, and reinforcing the risks and consequences of malpractice
• publishing classroom resources to support the safe and appropriate use of AI
• writing to exam boards to require stronger deterrence, detection and authentication measures
• working with the Department for Education to understand the implications of AI for assessment and to explore potential mitigations

We will continue to:

• monitor trends in AI-related misuse and use this evidence to inform our approach
• ensure our regulatory expectations are clear, proportionate and effectively implemented
• take action where misuse risks undermining the integrity of qualifications
• undertake research to understand how assessment design can mitigate the risks associated with AI misuse
• review the effectiveness of awarding organisations’ approaches to managing these risks

As GCSEs, A Levels and vocational qualifications undergo reform, Ofqual is also exploring ways to reduce the risk of AI-related malpractice while ensuring assessments remain appropriate to their purpose. This includes working with the Department for Education to identify appropriate mitigations and, where necessary, alternative approaches to assessment.

Next steps

AI presents significant opportunities to support innovation and improve the delivery of qualifications. At the same time, it introduces new and evolving risks that must be carefully managed to protect fairness, standards and public confidence.

Ofqual will continue to take a proportionate, evidence-based approach to the regulation of AI, supporting the safe and responsible adoption of new technologies while ensuring that qualifications remain valid, reliable and trusted.

Over the next 12 to 24 months, we will prioritise:

• reviewing and, where necessary, adapting regulatory processes and requirements
• engaging with the sector and, where appropriate, expanding our innovation support activity, including through our innovation service
• monitoring and further strengthening our approach to malpractice and risks arising from student misuse of AI

We will keep our approach under review as the use of AI in the sector develops and will report on our progress through our existing reporting arrangements.