AI Interviewer Instructions for Use
Compliance Considerations
As the employer, you are responsible for complying with the hiring, AI, privacy and recording consent laws that apply to you. This section explains how AI Interviewer is designed and highlights some laws you may need to consider in more detail with your legal team. It is not legal advice, and Ashby cannot advise on your specific legal obligations.
Candidate disclosures and consent
AI Interviewer is designed to tell candidates, before the interview:
- that the interview is conducted by AI
- that it will be recorded and transcribed (and video recorded, if you enable video)
- that the AI does not make the hiring decision
- how to request an alternative or accommodation
Candidates are asked to consent before recording starts. The AI disclosure is fixed text that can’t be edited, and the interviewer is designed to say at the start that it is an AI acting on your behalf. Please do not configure the interviewer, including its welcome message, to suggest that it is a person. You can tailor the rest of the terms, but the Product Specific Terms require you to keep the core disclosures listed above and the consent step.
Interview plans
You set the questions, criteria and minimum qualifications. AI Interviewer is designed to flag wording that may raise equal employment opportunity concerns, but this does not replace your own review, and proceeding past a flag is your decision and responsibility. You should:
- use minimum qualifications only for objective, job-related requirements
- avoid questions that local law restricts, such as criminal or salary history
- ask about work authorization only in terms of whether the candidate is authorized to work and will need sponsorship
- avoid questions about protected characteristics or other sensitive personal data, such as health, religion or family status, unless directly related to a genuine job requirement
If a candidate doesn’t meet a minimum qualification, the interview is designed to end early once the candidate confirms their answer. Ending the interview early does not reject the candidate, and a person should still review the interview and decide on next steps.
Human review
AI Interviewer is designed not to advance, reject or archive candidates automatically. Each completed interview creates a review task for a member of your team, who should review it and decide whether the candidate advances. Reviewers can see each competency signal with the observations and quotations behind it, the full transcript and the recording. To keep this review meaningful, we recommend that you:
- assign reviewers who understand the role and how AI Interviewer works, and who have authority to decide
- check signals against the transcript or recording, especially where a signal is negative or mixed, an answer is short, or transcription may have been affected by accent, speech patterns, disability or audio quality
- use AI Interviewer as one input among others, not as the sole basis for a decision
- offer an alternative process or accommodation to candidates who ask for one, and make sure recruiters know how to arrange it.
Integrity signals
Reviewers can see when a candidate switches browser tabs or takes more than 60 seconds to begin an answer. Candidates are told about these signals in the disclosure screens before the interview starts. The signals do not affect competency signals or trigger any action, and we recommend treating them as context, not evidence of misconduct, since a pause or tab switch can have innocent explanations, including disability or connection problems.
Biometrics
AI Interviewer is designed not to create voiceprints, faceprints or other biometric identifiers, and not to infer candidates’ emotions or analyze video for facial recognition. If you connect third-party tools that perform biometric processing, you are responsible for complying with the biometric privacy laws that apply, and we recommend reviewing those obligations before enabling them. If we add video analysis features in future, they will be off by default, and we’ll describe how they work before they can be turned on.
Bias testing
Many hiring laws prohibit selection tools that have a discriminatory effect, even where it is unintended. Ashby has AI Interviewer independently audited for bias, and a summary is available in the Ashby Trust Center. The audit uses test data, so it may not reflect your own interview plans, candidates or use of the tool, and it may not be enough on its own to meet local requirements. We recommend reviewing outcomes in your own hiring process over time.
Candidate data
You are the controller for candidate data, and access, correction and deletion requests should be handled through your privacy processes. You can delete individual evaluations and recordings at any time. For bulk deletion, contact your Ashby account team. Some laws require hiring records to be kept for a minimum period (four years in California), so we recommend you check with your legal team before deleting.
Jurisdiction-specific requirements
AI and hiring laws vary by location and can apply based on where your roles are located or where your candidates live. Below are some of the laws our customers most often ask about. This is not a complete list.
European Union
- Ashby treats AI Interviewer as a high-risk AI system under the EU AI Act, because it is used to evaluate candidates. Under the Act, Ashby is the “provider” of AI Interviewer, and you, as the organization using it in your hiring process, are the “deployer”. Each role has its own obligations.
- Requirements already in force: The requirement to tell candidates they are interacting with AI already applies, and AI Interviewer is designed to meet it. Deployers are also already required to take measures to support AI literacy among staff who use it. The ban on prohibited AI practices also applies, so neither AI Interviewer nor any tool connected to it should be used to infer candidates’ emotions.
- Requirements from December 2, 2027: Most other deployer obligations apply from this date. They include following these instructions, assigning trained human oversight, monitoring use, keeping logs, informing candidates and explaining decisions to candidates on request.
- GDPR: As the controller, you are responsible for the lawful basis, candidate notices and any data protection impact assessment, which is likely to be required. This guidance and the technical information in the Ashby Trust Center are designed to support your assessment.
- National laws: Some EU countries have their own rules on AI in the workplace. For example, Italy requires employers to inform workers and their representatives before using certain automated systems in hiring, and in other countries employers may need to consult works councils before introducing AI tools. We recommend checking with your legal team which national rules apply where you hire.
California
- Fair Employment and Housing Act regulations cover automated-decision systems, which can include tools that support hiring decisions, such as AI Interviewer.
- Discrimination: The regulations prohibit using these tools in a way that discriminates on a protected basis, and evidence of bias testing, or its absence, can be relevant if a claim is made. See Bias testing above.
- Record keeping: Employment records, including AI Interviewer data, must be kept for at least four years.
- Automated decision-making (CCPA): From January 1, 2027, if you are subject to the California Consumer Privacy Act, additional rules apply where automated tools make significant decisions, such as hiring, without meaningful human involvement. AI Interviewer is designed for human review, but that only counts if your reviewers understand the outputs, consider other information and have authority to make the final decision.
Illinois
- The Illinois Human Rights Act regulates employers’ use of AI in hiring.
- Notice: Employers must tell applicants when AI is used in hiring.
- Discrimination: The Act prohibits using AI in a way that discriminates on a protected basis, including using zip codes as a proxy. See Bias testing above.
- Video interviews: If you enable video for Illinois roles, we recommend following the Artificial Intelligence Video Interview Act, which covers notice, explanation, consent, limits on sharing, and deletion within 30 days on request.
New York City
- Local Law 144 regulates AI tools that play a major role in hiring and promotion decisions for New York City roles. Whether it applies depends on how you use AI Interviewer’s outputs. If it does:
- Bias audit: The law requires a bias audit less than a year old, with a summary published on your website. Ashby’s independent audit summary is in the Ashby Trust Center. It uses test data, which employers can rely on only if they don’t have enough historical data of their own. In that case, the published summary must explain why historical data was not used and how the test data was generated.
- Candidate notice: The law requires notice to New York City candidates at least 10 business days before use, describing that an automated tool will be used, the qualifications assessed and how to request an alternative process or accommodation.
Other jurisdictions
- Colorado (from January 1, 2027) and Connecticut (from October 1, 2027) will require notices when automated tools are used in employment decisions. Other jurisdictions are considering similar rules.
This guidance gives a general overview of some of the laws that may apply when you use AI Interviewer. It reflects the position at the date of publication and is not legal advice. Ashby can provide product and technical information to support your own assessment, but your legal advisers are best placed to advise on the requirements that apply where you hire.