Aipedia: Training providers and educators

AI in training and education: rules, risks and PI

Aipedia, by the Apex broking team · Last reviewed 2026-10-27

UK training providers can use AI for content, marking and learner support, but they stay responsible for the results, and Ofqual bars AI as a sole marker in regulated qualifications. Errors affect whole cohorts and can lead to claims from learners and clients.

Part of: AI in training providers and educators

In short

AI is used across training and education for course content, marking, assessment design, tutoring, proctoring, accessibility, admissions and coaching. There is no single regulator: the DfE guides schools and colleges, Ofsted inspects without assessing AI tools directly, Ofqual says AI may not be a sole marker in regulated qualifications, and JCQ guidance shapes malpractice policies. Unaccredited trainers have no regulator but owe reasonable care. There is no statutory PI requirement, and many trainers' PI policies cover the activity regardless of tools, but activity descriptions, AI clauses and data exclusions need checking.

How training providers use AI

AI can touch many parts of training and education. Providers may use generative AI chatbots and AI assistants to draft slides, workbooks, e-learning and question banks; to suggest marks and write feedback; to run AI tutors and enquiry chatbots; to proctor online exams; to caption, translate and simplify materials; to screen applicants and flag learners at risk; and, in coaching, to transcribe sessions and analyse 360 feedback. Learners use the same tools, which raises questions about the authenticity of coursework and portfolios.

The sector is broad. It includes schools and colleges, apprenticeship and skills providers, approved centres for regulated qualifications, professional exam preparation, corporate and compliance trainers, e-learning companies, and coaches and leadership developers. The rules that apply, and the claims that arise, differ across these groups.

The rules: who regulates AI in training

There is no single regulator, and commercial trainers delivering unaccredited courses have none. Where bodies do apply, their AI positions are:

DfE: the Generative AI in education guidance (updated 12 August 2025) for schools and colleges. Personal data is recommended not to be entered into generative AI tools; students' work must not be used to train AI without permission or a copyright exception; staff must check outputs; the professional and organisation remain responsible "regardless of the tools used"; pupil-facing use needs safeguards and KCSIE filtering and monitoring should cover generative AI. The product safety standards (updated 19 January 2026) apply to edtech suppliers.
Ofsted: does not inspect AI tools directly but looks at whether providers' AI decisions are sensible and serve learners, with data protection, safeguarding and bias considered within existing judgements. Its position is non-final.
Ofqual: its July 2026 approach says AI "may not be used as a sole marker" in regulated qualifications, students must not present AI-generated work as their own, sole AI remote invigilation is doubtful, and AI-generated items are under research.
JCQ: AI Use in Assessments, Revision 2 (30 April 2025) should inform centres' AI and malpractice policies.

Data protection law applies to every provider. Since 5 February 2026, UK GDPR Articles 22A to 22D govern significant decisions based solely on automated processing, such as automated admissions, progression or grading. For detail, see the Ofqual, Ofsted and DfE rules explained.

The main risks

Wrong content: AI-drafted materials with out of date law, invented references or missing safety steps, repeated across every cohort.
Unfair assessment: AI marking or AI-written questions that are inaccurate or biased, and AI proctoring that wrongly voids results.
Malpractice in both directions: accepting AI-written work, or wrongly accusing learners on the strength of a detector score.
Learner-facing AI: chatbots that give wrong answers or fail safeguarding expectations.
Data and IP: learner personal data and original work entered into AI tools without proper safeguards or permission.
Automated decisions and equality: scoring that excludes learners without human review, or AI formats that fail disabled learners.

We are not aware of any UK court decision, Ofqual enforcement action or Ofsted judgement specifically about a provider's AI use. The risks above follow from existing duties applied to new tools.

The PI position

No statutory PI requirement for training providers was found, and no regulator has made a statement on AI and PI. Clients, funders and awarding bodies may still require PI by contract. Many trainers' PI policies cover the professional activity regardless of the tool used, so a negligence claim arising from AI-assisted content, marking or advice will usually be considered on the normal terms. Gaps can arise from:

activity descriptions that do not reflect AI tutors, online assessment or certification work;
newer AI clauses or exclusions in some wordings;
data protection claims that sit under cyber rather than PI;
bodily injury exclusions, which matter for safety training;
contractual guarantees and indemnities beyond your legal duty;
aggregation of many claims from one repeated error.

Practical priorities

Map where AI is used across content, marking, learner support, data and decisions.
Keep a qualified person in charge of every output learners or clients rely on, and record the sign-off.
Never let AI be the sole marker in regulated qualifications, and keep people in significant decisions.
Write a learner AI policy and a fair malpractice process that does not rely on detector scores alone.
Keep identifiable learner data out of generative AI tools, and get permission before learners' work could train a model.
Update learner agreements, privacy notices and client contracts to describe AI use accurately.
Tell your insurer how you use AI and check how your PI and cyber policies fit together.

Aipedia guides for training providers and educators

Each guide below looks at one task or issue in detail:

Ofqual, Ofsted and DfE AI rules explained: what each body has published, its status, and which parts apply to commercial trainers.
AI-generated course materials and errors: responsibility for AI-drafted slides, workbooks and e-learning that learners rely on.
AI marking and feedback on learner work: Ofqual's sole marker rule, bias in AI marking, and appeals.
Learners using AI: malpractice and authenticity: JCQ and Ofqual expectations and liability for getting it wrong either way.
Relying on AI detection tools: false positives, fair process and discrimination risk.
AI tutors and learner-facing chatbots: safeguarding, wrong answers and supplier versus provider liability.
AI-built compliance and health and safety courses: outdated legal content and client reliance after incidents.
AI remote invigilation and proctoring: attribution, bias and disputes over voided results.
Learner data and IP in AI tools: DfE guidance on personal data and training models on learners' work.
AI-generated exam questions and assessments: validity, accuracy and bias in AI-written items.
AI in coaching and leadership development: 360 feedback, psychometrics, recordings and confidentiality.
AI in admissions and progression decisions: UK GDPR Articles 22A to 22D safeguards and discrimination risk.
AI translation, captions and accessibility: errors that disadvantage disabled or non-English-speaking learners.
Telling learners and clients you use AI: what to put in learner agreements and corporate contracts.
Does trainers' PI cover AI mistakes?: where cover usually responds, where it falls short and what to disclose.

How Apex helps

Apex arranges PI for training providers, educators, coaches and corporate trainers. A broker reviews your current cover, how it describes your activities, and how it treats AI-assisted content, marking and learner support, then explains any gaps in plain terms.

Sources

Frequently asked

Do training providers have to follow the DfE AI guidance?

The DfE guidance is aimed at schools and colleges. Other providers are not bound by it, but its principles, such as checking AI output, staying responsible for documents and keeping identifiable personal data out of generative AI, are a sensible benchmark for any educator.

Can AI mark learners' work?

AI can assist, but in regulated qualifications Ofqual says AI may not be used as a sole marker, and marking decisions must be accountable, explainable and open to challenge. In other settings, a person should still review AI-suggested marks before they affect learners.

Is PI insurance compulsory for training providers?

No statutory PI requirement for training providers was found. Clients, funders and awarding bodies may require it by contract. It is the main protection if a learner or client claims your training, assessment or advice was negligent and caused them loss.

Will my PI cover claims involving AI?

Usually the claim is considered on the normal policy terms, because many policies cover the professional activity regardless of tools. Check that your policy describes what you actually do, look for any AI clauses, and confirm how data and injury claims are treated.

Check your PI covers how you use AI

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 callback

Apex 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.