FCA authorised · FRN 724952 0117 325 0027Quote & buy →
Aipedia: Training providers and educators
Providers can use automated scoring in admissions and progression, but significant decisions made solely by a system need safeguards under UK GDPR Articles 22A to 22D. Learners must be able to get human review and contest the decision.
Part of: AI in training providers and educators
In short
Training providers may use automated scoring to screen applicants, decide progression and flag learners at risk. Since 5 February 2026, UK GDPR Articles 22A to 22D govern significant decisions based solely on automated processing. Article 22C requires safeguards: information about the decision, a chance to make representations, human intervention and a right to contest. In regulated qualifications, Ofqual says AI may not be a sole marker. Models trained on past cohorts can repeat bias and lead to discrimination complaints. PI may respond to negligent decision claims, subject to its terms.
Providers may use automated scoring at several points: screening applications for courses or apprenticeships, scoring entry tests, predicting which learners are at risk of dropping out, deciding who progresses to the next level, and allocating funded places. Some systems only advise staff. Others make the decision, for example automatically rejecting applicants below a score or withdrawing learners whose engagement data falls under a threshold. The difference between the two matters legally, and it is easy for an advisory score to become a decision in practice when busy staff simply follow it.
Section 80 of the Data (Use and Access) Act 2025 replaced the old automated decision-making rule with Articles 22A to 22D of the UK GDPR. It came into force on 5 February 2026 under the Commencement No. 6 Regulations (SI 2026/82). They govern significant decisions based solely on automated processing. Admissions, progression and automated grading with a significant effect on a learner are likely to fall within them.
Where such a decision is made, Article 22C requires the controller to put safeguards in place that enable the data subject to:
For regulated qualifications, Ofqual adds that AI may not be used as a sole marker, so automated grading in that context is not compliant in any case.
Automated scoring trained on past data can repeat past patterns. A model that predicts success from previous cohorts may disadvantage groups that were under-represented or less well served before, such as disabled learners, career changers or those with gaps in education. Ofsted notes that AI can repeat bias in its training or input data and looks at whether providers have considered that risk. Discrimination complaints can follow; see our note on the Equality Act 2010 disability provisions.
Illustrative scenario: a training provider uses an engagement score to decide which apprentices continue to the second year. Learners below the threshold are withdrawn automatically. One had been off sick with an agreed adjustment, which the system did not know. They complain that no person looked at their case and that the decision was discriminatory, and their employer questions the provider's process.
A claim that a provider negligently assessed or withdrew a learner, causing financial loss, may fall within a trainers' PI policy, subject to its terms. Data protection and discrimination allegations are treated differently across wordings, and some policies exclude them or limit cover to defence costs. See GDPR cover within PI and the Data Protection Act 2018.
Apex arranges PI for training providers and educators and can check how your policy treats data protection and discrimination allegations. See the training providers Aipedia hub, AI marking and feedback and learner data in AI tools.
If a rejection is a significant decision based solely on automated processing, UK GDPR Articles 22A to 22D apply from 5 February 2026. You must provide safeguards, including information about the decision, a way to make representations, human intervention and a way to contest it.
A person with the authority and information to change the decision must genuinely review it. A staff member who simply confirms the system's output without looking at the case is unlikely to be enough. Record what the reviewer considered and why they decided as they did.
No. Ofqual says AI may not be used as a sole marker in any assessment that is part of a regulated qualification. Expert human involvement and accountability for outcomes are required, and marking decisions must be capable of explanation and challenge.
Articles 22A to 22D are aimed at decisions based solely on automated processing. If a member of staff genuinely considers the case and makes the decision, the rules on solely automated decisions are less likely to apply, though general data protection duties and discrimination law still do.
Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.
Start a PI quote Request a callbackApex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.
