AI statement and instructions for use
Version of 12 November 2026
Lisa is an AI assistant for the first introduction with applicants. This page explains what Lisa does and does not do, how to use her responsibly, and how Lisa relates to the European AI Act and the GDPR. This is an English translation. If there is any difference, the Dutch version prevails.
1. What Lisa does
Lisa reads the vacancy and (if there is one) the CV, asks the candidate a number of questions in a chat, including the fixed questions the employer chooses, and summarises the conversation factually.
If the employer turns on scoring, Lisa also calculates how well the CV matches the vacancy (CV match) and gives a score for the answers. These are tools for the recruiter.
The underlying language model is from OpenAI (GPT-4o mini). Data is processed through the business API and is not used to train models.
2. What Lisa does not do
Lisa makes no decisions: she rejects no one and invites no one. That is always done by a human at the employer.
Lisa does not ask about age, ethnic origin, nationality, religion, health, disability, pregnancy, sexual orientation, marital status, political opinion or trade union membership, and does not take these into account in summaries or scores, even if the candidate mentions them.
Lisa does not recognise emotions and does not assess personality. No facial, voice or other biometric analysis is performed.
3. Transparency towards candidates
Lisa says in her first message that she is an AI assistant, and every conversation shows that the answers are shared with the employer (Article 50 AI Act, applicable since 2 August 2026).
Each vacancy has a privacy statement explaining what happens to the data, that a human decides, and that the candidate can ask for an assessment without Lisa or for an explanation.
4. How the AI Act applies to Lisa
AI systems for recruitment and selection, such as assessing and filtering applications, are high-risk under the AI Act (Annex III). Lisa falls into this category, especially when scoring is on.
The obligations for these high-risk systems apply from 2 December 2027 (postponed by Regulation (EU) 2026/1744). We are preparing with risk management, technical documentation, registration and logging, and will adapt Lisa and this page where needed.
The ban on emotion recognition in recruitment, the transparency obligation (point 3) and the AI literacy obligation already apply. Lisa is set up for this.
5. Using Lisa responsibly (for employers)
Never make a decision based solely on the summary or score. In case of doubt read the full conversation and the CV; they are available for every candidate.
Scoring is optional and found under Settings. If you use it, treat the CV match and score as a tool, not a verdict.
Do not set intake questions about the personal characteristics in point 2. Only ask what is needed for the job, such as a driving licence or availability.
Make sure colleagues who work with Lisa know this page (AI literacy). A short guide to go through with your team and have signed is at lisarecruit.com/ai/team. If a candidate asks for an assessment without Lisa or for an explanation, handle the application yourself or explain how the summary was produced.
Choose a suitable retention period under Settings. The Dutch Data Protection Authority advises keeping application data no longer than 4 weeks after the procedure, or up to 1 year with the candidate's consent.
Recruitment with AI often requires a data protection impact assessment (DPIA). The employer carries this out; we help with the information on this page and in the data processing agreement.
6. Limitations
A language model can make mistakes: a summary may miss or misrepresent something, and Lisa cannot read a CV saved as a photo or scan. Lisa translates answers given in another language, which may lose nuance.
A CV match says something about how the CV and vacancy compare, not about someone's suitability as a person. People without a 'perfect' CV can be excellent candidates.
7. Reports and questions
Do you see Lisa do something that is not right, such as a question that should not have been asked or an incorrect summary? Report it to info@lisarecruit.com. We look into every report and adjust Lisa where needed.