Prohibited use of artificial intelligence in AAT assessments

Last updated: 2 September 2024

The use of generative artificial intelligence (AI) offers opportunities for learning, qualification development and awarding, but it also poses a number of risks to the integrity of qualifications and assessments. As AI is an evolving rather than a new technology, regulations to govern its use in the development, delivery and award of qualifications are still being considered by regulatory authorities within the qualification sector.

In this policy we have focused on what is required under the current regulatory framework to ensure the authenticity of work completed by learners for assessments undertaken as part of AAT qualifications and End Point Assessments (EPAs).

This policy does not provide any guidance about how AI may or may not be used in learning or in the development of resources.

Scope and applicability

This policy applies to learners, training providers, individuals and any third-party involved in the delivery or award of AAT assessments and qualifications. A failure to follow the requirements of the policy may result in investigation and action being taken as considered appropriate.

This policy will require revision should the regulatory framework change or standards for the AAT qualifications and assessments be developed to include the use of AI.

Purpose and objectives

The aim of this policy is to prevent the inappropriate use of AI in AAT assessments and to ensure that assessment and qualification results for learners are valid, reliable and fair.

AAT’s qualifications are regulated by the Office of Qualifications and Examinations Regulation (Ofqual) in England, Qualifications Wales (QW), the Council for Curriculum, Examinations and Assessment (CCEA) in Northern Ireland, and Qualifications Scotland Accreditation.

AAT policy and procedures for development, delivery and award of assessments and qualifications, including End Point Assessments (EPA) must comply with the Conditions of Recognition (in England, Northern Ireland and Wales), the Regulatory Principles (in Scotland). As an End Point Assessment Organisation (EPAO), AAT must also comply with the Conditions for Apprenticeship Provider and Assessment Register (APAR).

Terms and definitions

  • AI: artificial intelligence or generative artificial intelligence.
  • AO: awarding organisation is an organisation that sets and awards regulated qualifications.
  • Assessor: a person who has the relevant subject matter expertise which they use to make assessment judgements and consistently differentiate between different levels of performance during marking.
  • Centre: an organisation that provides training to learners for assessments and/or acts as a venue to administer/manage assessments for learners.
  • EPA: End Point Assessment.
  • Invigilator (proctor): a person who supervises learners during assessments to ensure the assessments are conducted in accordance with the AO’s requirements and to prevent misconduct.
  • Learner: a student or an apprentice.

Policy details

Malpractice and maladministration

A breach of the principles included in this policy will be treated as malpractice or maladministration, however, the principles included in this policy are not intended to be an exhaustive list of what might constitute malpractice or maladministration. All allegations of malpractice or maladministration will be investigated to establish whether malpractice or maladministration has taken place.

Invigilation of assessments

Where AI is used to support invigilation of assessment, AI cannot be used as the sole invigilator for in-person or remote assessments. Invigilation for the entirety of in-person and remote assessments, including "record and review", must be conducted by a human invigilator.

Learners' assessment evidence

The use of AI is not permitted during AAT in-person or remote assessments. Evidence produced and submitted by learners for their assessments must be the learner’s own unaided work.

Centres are required to report all instances where they suspect or have evidence that the authenticity of assessment evidence submitted by or for a learner has been compromised by the use of AI.

For EPAs, AI must not be used to produce the report or portfolio evidence. The Institute for Apprenticeships and Technical Education (IfATE) works with employers to develop, approve, review and revise standards for apprenticeships, and IfATE has determined that where the use of AI forms part of the apprentice’s day-to-day work, it must be referenced within that work, fully referenced within the portfolio and completed in accordance with the assessment plan.  

Marking assessments

Marking schemes for computer marked tasks must be set by humans to mark fairly and consistently. The marking and moderation of all non-computer marked assessment tasks or assignments must be carried out in full by an appropriately qualified human-assessor.

AI may not be used to judge the performance of learners in any AAT assessment activity that contributes to the award of unit assessment results, a qualification or EPA.