What to Tell US Candidates Before Using an AI Notetaker

Tell every candidate, in writing and before the interview, that an AI notetaker will record and transcribe the conversation, why, who will see the notes, how long you keep them, and how to decline without penalty. Then get a clear yes. A notetaker that doesn't join the call shows candidates nothing, so your notice is the only notice they get.

That is the baseline everywhere in the US. On top of it, about 11 states require every party to consent to a recording, Illinois has three laws that can reach AI in interviews, New York City regulates AI tools that substantially drive hiring decisions, and federal anti-discrimination law still applies to whatever the notes are used for. This guide walks through each one and ends with a notice and a consent script you can adapt.

When a bot joins a Zoom, Teams or Google Meet interview, the candidate sees an extra participant, and the platform's own recording shows its own notice. A notetaker that captures audio on the interviewer's computer shows none of that. Nothing appears in the participant list, and the meeting platform cannot detect it. In an in-person interview there is no screen at all.

That is a good thing for the conversation: no bot staring back from the grid, no "Notetaker has joined" banner while a nervous candidate is trying to think. But recording laws are about recording people, not about how the audio is captured. With no bot and no platform banner, the whole disclosure becomes your job, so it has to be explicit, early and written.

Federal law only needs one party's consent, and you are a party. That is not the end of it. According to the Reporters Committee for Freedom of the Press, 11 states primarily require the consent of all parties: California, Delaware, Florida, Illinois, Maryland, Massachusetts, Michigan, Montana, New Hampshire, Pennsylvania and Washington. Oregon and Missouri require all parties for in-person conversations but one for phone calls; Connecticut and Nevada do the reverse. Hawaii and Maine require all parties only in particularly private places.

Interviews are remote, and candidates apply from anywhere. When the people on a call are in different states, the safe assumption is that the strictest rule applies. California's Supreme Court did exactly that in Kearney v. Salomon Smith Barney (2006), applying California's all-party rule to calls recorded in Georgia. Since you rarely know for sure where a remote candidate is sitting, the practical rule for hiring teams is simple: treat every interview as all-party consent.

• Ask in writing before the interview, with the confirmation email, not on the day.
• Ask again out loud at the start, and wait for a real answer.
• Treat silence or hesitation as a no.
• Ask every person on the call, including a second interviewer or a recruiter who joins late.
• If the candidate says no, record nothing, not even "just the audio for me".

Illinois is the state where AI in hiring is most regulated today, and it is also an all-party consent state. Three laws are worth checking for any role based there.

This 2020 law applies to employers that ask applicants to record video interviews and use AI to analyze those videos, for positions based in Illinois. Before the interview you must tell the applicant that AI may be used to analyze the video, explain how the AI works and what general types of characteristics it evaluates, and get their consent. You may share the video only with people whose expertise or technology is needed to evaluate the applicant, and you must delete it, including backups held by others, within 30 days of the applicant asking. Since 2022, employers that rely solely on AI analysis to decide who gets an in-person interview must also report demographic data to the state each year.

By its wording, the Act targets AI analysis of applicant videos. An audio notetaker on a live call is a different setup. But if your tool analyzes video, or rates how candidates come across on camera, read the Act closely, and use its notice and deletion rules as a model even where it may not strictly apply.

Since January 1, 2026, the Illinois Human Rights Act makes it a civil rights violation to use AI in recruitment, hiring, promotion and other employment decisions in a way that has a discriminatory effect, or to use zip codes as a proxy for protected classes. It is also a violation to use AI for those purposes without giving notice. The Illinois Department of Human Rights has authority to set when and how that notice is given, so check its current rules. If AI summaries or scores of interviews feed your hiring decisions, the cautious reading is that candidates should be told.

BIPA lists voiceprints as biometric identifiers. Before collecting one, a private company must tell the person in writing what it collects, why and for how long, get a written release, and publish a retention schedule with destruction guidelines. Individuals can sue, with liquidated damages of $1,000 per negligent violation and $5,000 per intentional or reckless one.

NYC Local Law 144 covers automated employment decision tools (AEDTs). In the city's own FAQ, an AEDT is a computer-based tool that uses machine learning, statistical modeling, data analytics or AI, helps make employment decisions, and substantially assists or replaces discretionary decision-making. The rules count a tool as substantially assisting when its simplified output, such as a score or a ranking, is the only factor, is weighted more than any other criterion, or is used to overrule human judgment.

A transcript on its own is not a score. But many notetakers now rate candidates, rank answers or suggest a hire or no-hire. If your team relies on that output in one of those three ways for a role located in NYC, the law requires a bias audit within one year before use, a public summary of the results, and a notice to candidates who live in NYC at least 10 business days before use. The notice can go in the job posting, by mail or email, or on the careers section of your website. It must say that an AEDT is used, which job qualifications or characteristics it assesses, and how to request a reasonable accommodation.

Enforcement is likely to tighten. In December 2025, the New York State Comptroller's audit called the city's enforcement of the law ineffective, and the city agency agreed to implement most of the recommendations. The simplest compliance posture for an interview notetaker is to use its output as notes a person reads, never as the decision.

In January 2025 the EEOC removed its technical assistance on AI in hiring (a May 2022 document on the ADA and a May 2023 document on Title VII) from its website. The laws themselves did not change. Title VII still prohibits selection procedures with an unjustified disparate impact on a protected group, and the ADA still requires reasonable accommodation for applicants with disabilities, whether a person or software is involved.

• Offer an accommodation route in the notice: a candidate with a speech or hearing disability may need a different format.
• Do not judge communication skills from transcript quality. Transcription is less accurate for some accents and speech patterns, and that is the tool's limit, not the candidate's.
• Keep protected information out of the notes: age, family status, health, religion, national origin.
• Score against job criteria and keep a person accountable for every decision.

California's Civil Rights Council regulations on automated-decision systems in employment took effect on October 1, 2025, and include keeping related records for four years. The California Privacy Protection Agency's rules on automated decision-making technology require covered businesses to comply by January 1, 2027. Colorado replaced its 2024 AI Act with SB 26-189, signed on May 14, 2026, and effective January 1, 2027. If you hire in these states, add them to your counsel's review.

Send this with the interview confirmation, so the candidate has time to read it and say no without pressure. Replace everything in brackets, and remove any line that is not true for your tool.

• Illinois roles: if AI analysis informs your decision, add a plain description of how it works and what it evaluates, and give notice under the Human Rights Act.
• NYC roles where an AI score substantially assists the decision: send the notice at least 10 business days before use, name the qualifications assessed, and link your bias audit summary.
• If the tool builds voiceprints and the candidate is in Illinois: get a separate written release under BIPA before the interview.

Then stop talking and wait. Start recording only after a clear yes. If the candidate hesitates, say "no problem" and take notes by hand. A candidate who feels they had a real choice is a candidate who trusts your process.

1. Map where your roles, interviewers and candidates are, and default to all-party consent.
2. Ask your vendor in writing: where audio is processed and stored, whether it trains models on your data, and whether it builds voiceprints.
3. Decide whether the tool only takes notes or also scores candidates. If it scores, check NYC Local Law 144 and the Illinois laws.
4. Send the written notice with the interview confirmation and record the answer.
5. Ask again at the start of the interview, and honor a no with handwritten notes.
6. Limit access to the hiring team, and set a retention period you actually follow.
7. Delete on request and on schedule, and keep people accountable for every decision.

Reline is a notetaker that doesn't join the interview. It captures your microphone and your computer's audio locally, on Mac, Windows or Linux (beta), so no bot appears in the call. Capture is the only local step: transcription, AI notes and storage run in our cloud, which belongs in your candidate notice. Reline separates your side from the other side by audio source, and it does not build a saved voiceprint to recognize a person across recordings.

Notes are private by default: a workspace role alone grants no access, and each viewer needs an explicit grant, so candidate notes stay with the hiring team. The Candidate Scorecard Lens scores only what was discussed and marks the rest N/A, but it also drafts scores and a hiring recommendation, so treat its output as notes a person reviews; if anyone starts treating its scores as the decision, that is when NYC Local Law 144 matters. Reline does not delete notes on a schedule by itself, so your retention period is a team rule you follow. To honor a deletion request, delete the note and then delete it permanently from the trash (trashed notes are otherwise purged after 30 days), which also removes its recording. Reline does not send the notice for you. The script above is your job, and it is the part that matters most.