Data last verified: 12 October 2026. Figures re-checked against their primary sources under the Whito rolling re-verification programme.
Last Updated on October 12, 2026
Status: specified, not yet collected. This page sets out a study Whito has designed and will run. It contains no findings. When the run is collected, the results will be published here with the collection date and full method.
Apprenticeship delivery in England runs on a hard register, APAR, published and downloadable from the Department for Education, and qualifications run on Ofqual’s. Both are the kind of yes-or-no public record AI answers should be easy to check against, and almost nobody ever does. This study does, for every recommended name.
The question we will ask
“Recommend an apprenticeship training provider in [town]”, with the town name substituted for each of the twelve.
Engines: ChatGPT, Gemini and Google AI Mode, asked the same question on the same day, each in a fresh session.
Where
Twelve UK towns and cities, the same set Whito uses across its register studies: Leeds, Manchester, Birmingham, Bristol, Sheffield, Nottingham, Leicester, Coventry, Newcastle, Cardiff, Southampton and Brighton.
The register, and how every name gets classed
Every recommended provider goes against the APAR download on the day of collection, plus Companies House for entity status, and any named qualification in an answer gets checked on the Ofqual register. Each name lands in exactly one class:
- On APAR. Entitled to deliver funded apprenticeship training, as recommended.
- Real provider, not on APAR. A genuine training business recommended for a funded route it cannot deliver.
- Unmatchable. No traceable entity behind the recommended name.
- Gone. Dissolved companies and providers removed from the register, still being recommended.
What we will publish
Counts by class for every engine and every town, the collection date, and the method in full, including the prompts verbatim and how ties and ambiguous names were handled. England only, since APAR is the English apprenticeship register.
Why this check, in this trade
An employer who starts an apprenticeship conversation with an off-register provider loses weeks, and a learner can lose a funded place. The register that prevents it is a downloadable spreadsheet. The study measures how often the engines’ confident recommendations would survive a VLOOKUP.
ed, and gave a registration number twice. https://whito.co.uk/gas-engineers/research/ai-gas-engineer-recommendations-gas-safe/. Free to quote with a link.Corrections and right of reply
If your business is named in this research and you think we have a fact wrong, email hello@whito.co.uk. We check it against the register, correct it within five working days if we are wrong, and publish your statement alongside if you want us to. Full route: corrections, right of reply and removals.
Questions
Can I trust an AI answer that says a gas engineer is Gas Safe registered?
Treat it as a claim to check. On 3 October 2026, 36 answers from ChatGPT, Gemini and Google AI Mode in 12
If your business is named here and you think we have a fact wrong, email hello@whito.co.uk and we will check it and correct it. If you want to respond, send us a statement and we will publish it alongside. Full detail: Corrections, right of reply and removals.
