COMPLIANCE · AWARDING BODIES
Can Awarding Bodies Approve AI Marking Tools? What Business Qualification Providers Are Saying About AI
Out loud the question is is this allowed. Underneath it’s will my awarding organisation back me up if something goes wrong. How to get a straight answer to both.
By Eduface · July 2026 · 8 min read
There’s a version of this question that training providers ask out loud, and a slightly different version they’re actually worried about underneath it. Out loud: is this allowed. Underneath: if I adopt this and something goes wrong, will my awarding organisation back me up, or will I be the one explaining myself. Those are different problems, and the second one is the one that actually keeps directors up at night.
Why the answer genuinely varies
If you’ve asked two different awarding organisations about AI marking and gotten two different answers, that’s not a sign either one is wrong. Ofqual’s own regulatory framework leaves room for awarding organisations to set their own specific policies within the broader principle that human judgement has to remain part of any marking decision, and Ofqual has said explicitly that differences between awarding organisations are expected. Some have published detailed guidance already. Others are still working it out. Neither position is a red flag on its own, it mostly reflects where each organisation happens to be in its own process, and how directly AI marking touches the specific qualifications they regulate.
What awarding bodies are actually weighing up
Behind the scenes, awarding organisations are wrestling with the same tension Ofqual identified in its own engagement with the sector: validity, reliability, and fairness on one side, against the genuine appeal of speed and lower marking cost on the other. That’s not a hypothetical trade-off dreamed up by regulators. It’s the actual conversation happening inside these organisations right now, and it’s worth knowing that when you’re the one asking them for a position, because it explains why the answer sometimes takes longer than you’d like.
There’s also a legitimate accountability question sitting underneath all of this that goes beyond education specifically. Danielle Citron and Frank Pasquale’s well known argument about automated scoring systems, made originally about credit and employment decisions rather than education, is that these systems tend to be adopted faster than the oversight structures needed to make them accountable, precisely because they look objective and efficient on the surface (Citron & Pasquale, 2014). Awarding organisations approaching AI marking cautiously are, in effect, trying to avoid exactly that gap opening up under their own qualifications.
What actually helps you get a straight answer
Ask a specific question, not a general one. Are we allowed to use AI invites a cautious, hedged answer. Here is exactly how our tool works: the AI drafts a grade with reasoning, an assessor reviews it, and nothing reaches a learner without explicit assessor approval, does this meet your requirements invites a much more useful one, because you’ve done the work of showing where the human decision actually sits.
Get the answer in writing. Not because anyone’s assuming bad faith, but because a verbal assurance from a call six months ago is not something you want to be relying on if a decision is ever challenged.
Ask what changes when the position updates. Ofqual has said clearly that this whole area is still developing and that further guidance is coming as evidence builds. Your awarding organisation’s current position is a snapshot, and it’s worth knowing upfront whether they’ll notify you directly when it shifts or whether it’s on you to keep checking.
Bring documentation, not just a description. If you can show an actual workflow diagram, or a short written policy for how grades move from AI draft to assessor approval, that’s a much stronger starting point for a conversation than describing it verbally and hoping it lands the way you intended.
The framing that tends to land well
Providers who get further, faster with this conversation tend to lead with the same thing: they’re not asking permission to remove human judgement from marking, they’re asking for confirmation that a specific, documented, human-approved workflow meets the bar. That’s a genuinely different conversation than the one an awarding organisation is bracing for when a provider calls asking about using AI to mark. Framing it accurately from the start tends to shorten the whole process considerably.
Frequently asked questions
Why do different awarding organisations give different answers to the same question?
Because Ofqual’s framework deliberately leaves room for that, and awarding organisations are at different stages of publishing formal guidance. Variation is expected, not a sign that one of them is getting it wrong.
What should I actually ask my awarding organisation, word for word?
Describe your specific workflow, including exactly where and how an assessor reviews and approves each grade, and ask directly whether that workflow meets their requirements. A specific question gets a more useful answer than a general one.
Should I get their answer in writing?
Yes. Not out of suspicion, but because a policy position needs to be something you can point back to later, not something you remember from a phone call.
What happens if my awarding organisation hasn’t published a position at all?
See our companion piece on what Ofqual’s guidance means for awarding organisations and centres. Silence isn’t permission, and it isn’t prohibition either, it just means you need to ask directly.
Sources
Ofqual. Ofqual’s Approach to Regulating the Use of Artificial Intelligence in the Qualifications Sector. gov.uk.
Citron, D. K., & Pasquale, F. (2014). The Scored Society: Due Process for Automated Predictions. Washington Law Review, 89, 1-33.
Walk into that conversation with the workflow ready
Eduface gives you a documented, human-approved marking workflow to put in front of any awarding organisation. Book a demo or start free.