When AI interviews for you: the evidence questions in 24-hour recruitment

An AI that interviews when you sleep

A recruitment AI developed in Sweden now conducts interviews around the clock. Tengai AB reports that 40 per cent of AI interviews in Sweden take place outside working hours, between 5 pm and 8 am.^[1] The AI avatar asks questions, listens to answers and writes a structured report for the hiring team to review.

The company says the system allows candidates to choose when they are interviewed, and removes bias by asking everyone the same questions. The recruitment team controls which questions are asked and makes the final decision.^[1]

That division of labour raises a set of evidence questions. When the AI talks to candidates on behalf of the hiring organisation, who keeps the record of what was asked and answered, and can the candidate get it later?

What the law says about recruitment AI

AI used to evaluate people for recruitment counts as high risk under Annex III of the AI Act.^[2] From 2 December 2027, such systems must log events automatically over their lifetime, for traceability and monitoring.^[5] The law says what to log. It does not say how the logs must be protected, but a log the operator can quietly edit is easy to challenge when someone questions a decision.

Under Article 26.11, a person affected by a decision taken with the support of high-risk AI has the right to a clear and meaningful explanation of the AI system's role in the decision-making procedure and the main elements of the decision taken.^[3] That right applies from 2 December 2027.

Article 26.7 adds a separate duty: before high-risk AI is used at the workplace, employees and their representatives must be informed and consulted.^[4] It does not just apply to the candidates being interviewed. The people working for the organisation using the AI have a right to know.

What this means for you

If you use AI to interview or assess candidates, ask these questions now:

• Who keeps the record of each interview: the AI provider, your organisation or both?
• Could you show a candidate, six months later, exactly what the AI asked them and what they said?
• If a candidate asks for an explanation under Article 86, does your contract with the provider give you the information you need to provide it?
• Have you informed your employees and their representatives that you are using high-risk AI in recruitment, as Article 26.7 requires?

The law asks for the logs. Whether they hold up when questioned is another matter.

Making interview records checkable

Here is where a digital witness fits. The thing to seal is each interview record as it is written: the candidate reference, the questions asked, the AI's assessment and the timestamp. Sealing takes a fingerprint of the record, not the content, so the interview itself never leaves your systems.

If a decision is challenged months later, anyone you share the record with can check that it is exactly what was written on the day. The check works even if the person doing it does not trust your organisation, because the sealed fingerprint sits on a chain where any later change would show, even one made by us.

A digital witness does not show that the questions were fair, that the assessment was accurate or that the final decision was lawful. It shows that the record you are relying on has not changed since it was sealed. You can build it into your recruitment workflow with WitniumChain through the REST API, the SDK or the MCP server. The content stays with you. The first 2,000 witnesses a month are free. Read the docs.

How it would work

1. The AI writes the interview record to your system.
2. Your system sends a fingerprint of the record to WitniumChain and gets a witness ID back.
3. Later, when the record is questioned, you share it with the candidate or their representative. They calculate the fingerprint again and compare it with the sealed one.

What to watch

The high-risk requirements for Annex III systems, including recruitment AI, apply from 2 December 2027. Article 26.7, on informing employees before high-risk AI is used at the workplace, applies from the same date.^[2] Article 86, the right to an explanation, also takes effect on 2 December 2027.^[3]

If the AI learns from the interviews it conducts, the data governance requirements in Article 10 will also apply.^[6] That means documenting where the training data came from and whether it was representative.

The Commission is expected to publish detailed guidance on the high-risk obligations during 2027. In the meantime, the questions above are a reasonable place to start.