AI Transparency and Disclaimer Notice

When you ask for a report, we send the model all of your assessment answers, including free text, your calculated scores, and the language you use the site in, so the report is written in that language. For the follow-up roadmap we also send your follow-up answers and the first report from the same attempt. If a university gives you access, we tell the model which country it is in, so its suggestions fit where you study.

We do not send your email address, your name, your payment details, your account identifier, your IP address, the name of your university, or reports from your other attempts.

Your answers pass through a sanitiser first, which strips content that could be used to manipulate the model. That is a security measure. Do not rely on it to remove personal information — it is not designed for that and will not catch everything.

We do not train AI models on your answers. Not ours, not OpenAI’s, not anyone’s. If we ever decide we want to, we will come and ask you first, rather than treating something you agreed to in the past as permission. Our Privacy Policy covers the contractual position and where the processing happens.

There is a second, separate use of AI here worth knowing about. We also use AI to read across many people’s answers at once — summarising which questions confuse people, where the scoring looks wrong, and where reports come out weak — so that we can fix those things. That analysis runs on data with the identifying details stripped out, it produces themes and figures rather than anything about one person, and it is still not training: the model learns nothing and keeps nothing.

Two things follow from that. A change we make because of it — a reworded question, a corrected weight — is a human decision informed by the analysis, not something a model decided on its own. And you can tell us to leave your data out of it entirely, without losing anything: how it works and how to object.

• It invents things. Language models produce text that reads convincingly whether or not it is true. A report can state a salary, a qualification, an entry requirement, a professional body or a course that does not exist — in exactly the same confident tone as the parts that are correct. You cannot tell them apart by reading.
• It is out of date. The model was trained on information up to a point in the past and has no live connection to the labour market. Salaries, entry requirements, apprenticeship routes, visa rules and funding all change, and it will not know.
• It does not know your local job market — where you live, what employers are near you, or what is being hired for in your area. Its knowledge of education systems, qualifications and pay is uneven between countries, and it may describe the routes of one country as if they applied in another.
• It knows nothing about you beyond what you typed. It cannot see your grades, your finances, your health, your caring responsibilities or your immigration status unless you wrote them down.
• It can reflect bias in its training data, including assumptions about who does which job. If a report seems to steer you by reference to something that should be irrelevant, that is a fault and we want to hear about it.
• It is not consistent. Ask twice and you will get two different reports. Neither is the “real” answer.
• No human reads it before you do. The report goes from the model to your screen. Nobody here checks it, and no qualified careers adviser reviews or approves it.

• Not professional careers advice or counselling. We are not a regulated careers service.
• Not a psychometric test. The questionnaire is not validated, not normed against any population, and has no published reliability or validity data. It must not be used for recruitment, selection or admissions.
• Not medical, psychological or mental health advice, and not a diagnosis of anything.
• Not legal, financial, tax, immigration or admissions advice. Anything a report says about visas, loans, funding, entry requirements or professional registration must be checked against an official source.
• Not a prediction and not a guarantee. No report promises you a job, a place, a salary, or a career you will enjoy.

The service produces information for you to consider. It does not decide whether you get a job, a place on a course, a loan or anything else, and it is not connected to any organisation that makes those decisions.

Your scores and reports are not shared with any university, employer, school or careers service. Nobody sees them unless you choose to show them. If a university gives you access, it sees only anonymous totals for all its students together, never your results. Our contract with it forbids using ADAQNOto decide anything about a student — admission, grading, discipline or anything else.

Because nothing here produces a legal or similarly significant effect on you, we do not consider Article 22 of the GDPR — the right not to be subject to a solely automated decision — to be engaged. We are showing you our reasoning rather than asking you to take it on trust, and our Privacy Policy sets it out more fully.

Whatever the strict legal position, you can always ask us to look at a result. We offer this because it is right, not because we have concluded we must.

• Ask us to explain the logic of the scoring model in plain English.
• Ask a person here to look at your report or scores and comment on it.
• Tell us you think a result is wrong or unfair, and have us review it.
• Ask us to delete a report you would rather did not exist.

Email privacy@adaqno.com. We aim to reply within 3 working days and to complete a review within one month.

A questionnaire that asks what you have failed at and what you are afraid of will sometimes receive answers describing real distress. So that the service does not respond to that by cheerfully listing job clusters, your free-text answers are checked against a fixed list of phrases, in English and in Spanish, suggesting self-harm or suicidal thoughts.

If one matches, support information is shown above your report and the model is instructed to write more carefully. Your report is not blocked, delayed or changed in its career content — you asked for it and you are entitled to it.

What this is not:

• It is a keyword match, not an assessment. It will miss things and will sometimes fire on ordinary phrasing.
• Nothing is stored. The result exists for one request and is then gone — not saved, not attached to your account, not visible to anyone.
• Nobody is alerted. No human is notified, and we do not monitor individuals.
• It is not a safety net. Please do not rely on it.

The support information shown is for your country: the country of your partner university if you have one, otherwise the approximate country of your connection, and an international directory of helplines where we have no local list.

Please tell us. Email support@adaqno.com with what the report said and what was wrong with it.

This is the single most useful thing you can do for the product. We cannot see your reports as a matter of routine, and a failure mode we do not know about is one we cannot fix. We particularly want to hear about invented facts stated confidently, advice that would be harmful to follow, anything that reads as biased or steering, and anything distressing.

European Union. We are based in Hungary and offer the service across the EU, so the EU AI Act applies to it. Our reading is that it is not a “high-risk” system: it is a self-help tool, and it is not used to decide admission to education, to evaluate learning, to assign a level of education, to monitor students during tests, or to recruit or select anyone — and our contracts with universities forbid them from using it for any of those. What applies to us are the Act’s transparency duties for AI-generated content. Your report is labelled as an AI report wherever it appears, and this notice explains what that means.

Elsewhere. Many countries, the United Kingdom among them, have no general AI statute. There our obligations come from data protection law, consumer law, and the ordinary law of negligence and misrepresentation. Wherever you use the service, the same notice and the same limits apply. If a law where we offer the service requires more, this page will change with it and we will say so.

If we change the AI provider, the model, or what we send it, we will update this notice and change the date at the top. If the change materially affects what the service does with your answers, we will email you.