Notice for candidates
It applies to live pre-screening; the public demo is covered by the general privacy notice.
The essentials
- A “No” answer to a knock-out question closes your application for that position automatically, without human involvement. A person can review this if you ask.
- If you complete the pre-screening, AI scores your two open answers. The system doesn’t reject anyone based on the score: a person at the company decides on invitations and rejections. Red-rated applications must be reviewed too, and we log this.
- Your open answers are processed by the AI provider Anthropic in the USA, without your name or email address.
- We keep your data and the review log for 6 months after the position closes.
1. Who handles your data
Controller: Prime AI Tech Consulting LLC, 1621 Central Ave, Cheyenne, WY 82001, USA. Email: info@primeaitechconsulting.com. The controller’s establishment in the EU: Ősz utca 12, 2040 Budaörs, Hungary. The company’s manager, Papay Miklós, can be reached there, and by email at the same address. The processing is carried out in the context of this establishment, so the GDPR applies directly and the competent supervisory authority is the Hungarian National Authority for Data Protection and Freedom of Information.
The company advertising the position is an independent controller responsible for the hiring decision. It processes your application as a step prior to entering into a contract, at your request, under Art. 6(1)(b) GDPR. The employer is responsible for the review. It informs you about its own processing.
2. What data we process, why and for how long
| Purpose | Data | Legal basis (GDPR) | Retention |
|---|---|---|---|
| Running the pre-screening | Name, email address, the position, your answers to the knock-out questions, skills test, situational task and survey | Legitimate interest, Art. 6(1)(f): orderly assessment of applications against uniform criteria | 6 months after the position closes |
| Assessing the application | Scores, green, yellow or red rating, reasons per criterion, quotes from your answers, any warning | Legitimate interest, Art. 6(1)(f): fast and uniform pre-assessment of a large number of applications | 6 months after the position closes |
| Automatic exclusion by knock-out question | The “No” answer, the fact and time of the exclusion | Legitimate interest, Art. 6(1)(f): filtering out applications that don’t meet the basic requirements of the job. The automated decision is permitted by Art. 22(2)(a), as it is necessary for entering into the employment contract with the employer. | 6 months after the position closes |
| Decision and notification | The decision of the company’s employee, interview time, notifications sent to you | Legitimate interest, Art. 6(1)(f): recording the decision and notifying you | 6 months after the position closes |
| Review log | Your application’s identification number, and who made what decision and when | Legitimate interest, Art. 6(1)(f): evidencing human review | 6 months after the position closes |
| Security and operations | IP address, technical logs | Legitimate interest, Art. 6(1)(f): system security and preventing abuse | Up to 30 days |
If a complaint or legal dispute is pending, we keep the data concerned until it is closed. We use your data for other positions only with your separate consent. Providing the data is not a statutory or contractual obligation, but it is needed for the application: we can’t process your application without it.
3. Automated decisions and AI scoring
Knock-out questions: an automated decision
Knock-out questions cover the requirements of the job, such as shifts or a licence. If you answer “No” to any of them, the system immediately closes your application for this position and you can’t continue. This is a fixed rule, not an AI decision. No person reviews it beforehand: an employee only sees it in a spot check or at your request. As a result you won’t be assessed for this position or invited, unless a review changes the decision.
You have the right to obtain human intervention, to express your point of view and to contest the decision. Write to info@primeaitechconsulting.com with the position and your email address. An employee will look at your application and can change the decision. We reply within one month at the latest, but it’s worth writing soon, as the position may be filled in the meantime.
AI scoring: a recommendation for a human decision
Anthropic’s Claude model scores your two open answers against the criteria set for the job. For each criterion it records whether the answer meets it, partly meets it or doesn’t, and writes a short quote, reason and summary. It can also flag an answer that knowingly commits to breaking a safety, legal or ethical rule.
The total score has three parts: skills test (share of correct answers), situation (75% open answers, 25% the multiple-choice question) and survey (statements about yourself). The company sets the weights and colour thresholds. If the AI is unavailable, rule-based scoring assesses the open answers, using keywords and the length of the answer.
The rating alone doesn’t reject anyone. A person at the company can see your answers, scores and reasons, and decides on an invitation or rejection. They must review red-rated applications too, possibly as a group. They confirm the review with a tick, and we log who made what decision and when under your application’s identification number. The order and colour do, however, influence whom they look at first.
We don’t analyse faces, voices or emotions. The AI is instructed to assess only the content of the answer, not spelling or style, and not to infer your gender, age, origin, health or any other protected characteristic.
4. Who receives your data
- The company advertising the position: its HR staff see your name, email address, answers, scores and reasons.
- For staffing, the client: a profile without name or contact details, with initials, scores and a summary.
- Anthropic, PBC (USA): AI scoring, as a processor. It receives your two open answers with the task and criteria, without your name or email address. The content of your answers can still point to you, so this is pseudonymised, not anonymous, data. Under its commercial terms it doesn’t train its models on data sent through the API and deletes it within 30 days. It may keep it longer where the law requires or to enforce its usage policy (up to 2 years for flagged content).
- Vercel Inc. (USA): hosting and operations.
- Resend Inc. (USA): email notifications.
Please don’t put personal data unrelated to the job in your open answers, especially about your health or other people.
5. Transfers to the USA
Transfers to the USA rely on the EU–US Data Privacy Framework where the recipient is certified, and otherwise on the European Commission’s standard contractual clauses. You can request a copy at info@primeaitechconsulting.com.
6. Your rights
- Access: you can ask what data we hold about you, request a copy of it, and we’ll explain in plain terms which of your data and which criteria shaped your assessment.
- Rectification, erasure and restriction of processing.
- Objection: you can object, on grounds relating to your situation, to processing based on legitimate interest.
- Withdrawing consent: where processing is based on your consent (for example for another position), you can withdraw it at any time. This doesn’t affect the lawfulness of earlier processing.
- For automated decisions: human intervention, expressing your point of view and contesting the decision (see section 3).
Send requests to info@primeaitechconsulting.com. We reply within one month at the latest. You can complain to the Hungarian National Authority for Data Protection and Freedom of Information (naih.hu) or to the EU supervisory authority of your habitual residence, place of work or the place of the alleged infringement. You can also go to court: in Hungary you can bring the case before the regional court (törvényszék) of your home or place of residence.