AI interviewerrankingsvendor evaluationbias auditvideo interview

AI interviewers in 2026: 28 tools that interview candidates, ranked on evidence

Twenty-eight AI interviewer tools, ranked by their HireAIScore score: what each one scores candidates on, what its bias audit actually tested, and what to ask before you let an AI run first-round interviews.

By HireAIScore editorial · Updated · 16 min read

On this page 19 sections
  1. What counts as an AI interviewer
  2. How we ranked them
  3. AI interviewers ranked at a glance
  4. Most AI interviewer bias audits never looked at a real applicant
  5. 1. Eightfold AI - 67/100
  6. 2. HireVue - 62/100
  7. 3. HeyMilo - 60/100
  8. 4. Phenom - 59/100
  9. 5. Apriora (Alex) - 59/100
  10. 6. Braintrust AIR - 59/100
  11. 7. Fountain - 58/100
  12. 8. Sense - 57/100
  13. 9-13. Sapia.ai, Tenzo AI, Classet, HackerRank, Glider AI
  14. The rest of the field
  15. What AI interviewers score candidates on
  16. The rules that apply to AI interviews
  17. How to choose an AI interviewer
  18. If you're the candidate
  19. Frequently asked questions

Key takeaways

  • Eightfold AI ranks first at 67/100, ahead of HireVue (62) and HeyMilo (60). They are the only three AI interviewer vendors to reach a D; no tool in the set reaches a C.
  • Of the thirteen highest-scoring vendors, seven audit their interviewer's scores on test or synthetic data rather than real applicants, and four publish no current audit that covers the interviewer at all.
  • Among those thirteen, only HireVue's and Classet's audits use real candidate data, and HireVue's are posted by its customers rather than by HireVue itself.
  • Most tools score the transcript against a rubric, but several also score English proficiency, speech delivery or AI-assistance signals, or watch head and eye movement for cheating. Ask for a list of every score the tool produces.
  • Ask any vendor what data its bias audit used, which of its scores the audit covers, and whether the tool can reject or advance a candidate without a person in the loop.

An AI interviewer is software that runs the interview itself. It phones the candidate, or meets them on video as an avatar, or chats with them by text; it asks the questions, follows up, and hands the hiring team a score, a transcript and a recording. In 2026 they stopped being a niche: HireVue launched one in June, HackerRank's went generally available in October, and Eightfold, Phenom and Sense all sell one alongside their core products.

We score every vendor in the directory on the same rubric, from public evidence. Below are the 28 that sell an AI interviewer to employers, ranked by score, with what each interviewer scores candidates on and what its bias audit actually tested.

What counts as an AI interviewer

The tools here come in three formats:

  • Voice agents that phone or call candidates: Fountain's Anna, Classet's Joy, Tenzo AI's Morgan, ConverzAI's Virtual Recruiter, Braintrust AIR, BrightHire Screen, HireVue's AI Interviewer and HackerRank's Chakra.
  • Video or avatar interviewers that hold a live two-way conversation: Apriora's Alex, Tengai's browser avatar and Interviewer.AI's conversational interviewer. Several voice agents, such as HeyMilo and Ribbon, also run video.
  • Text chat: Sapia.ai's Chat Interview, five to seven written questions with no video or audio.

Two neighbouring kinds of tool are not in this ranking. One-way video platforms, where candidates record answers to fixed questions, are covered in our HireVue alternatives ranking. AI note-takers that record interviews people run are not interviewers. We also left out VidCruiter, whose voice AI Interviewer is listed as coming soon, and Mercor, whose help centre says it does not offer its AI interviewer to other platforms.

How we ranked them

Each vendor gets a 0-100 score in seven categories, weighted into one overall score:

CategoryWeightWhat it asks
Article 11 technical documentation20%Annex IV-style documentation of how the system works
Bias audit transparency18%Published audits, methodology, auditor, cadence
FRIA support15%Material a deployer needs for an impact assessment
Data governance disclosure15%Training data, retention, subprocessors
Human oversight design12%Review, override and audit-trail controls
Post-market monitoring12%Change logs, incident handling, ongoing testing
Customer documentation8%Deployer guidance, notices, instructions for use

One caveat matters more here than in most rankings: the score rates the vendor, not the interviewer. Eightfold's 67 reflects documentation and audits across its whole talent-intelligence platform, of which the AI Interviewer is one product. Where a vendor's evidence for the interviewer itself is thinner than for its other products, the write-up says so.

Only three vendors reach a D. The rest are F, so read the ranking as "best-evidenced first", not as a list of endorsements.

AI interviewers ranked at a glance

RankToolFormatScoreLast reviewed
1Eightfold AI - AI InterviewerInterview agent67 (D)2026-11-14
2HireVue - AI InterviewerVoice62 (D)2026-11-14
3HeyMiloVoice, phone, video60 (D)2026-11-14
4Phenom - voice screening agent, AI InterviewerVoice59 (F)2026-11-14
5Apriora - AlexVideo, phone, SMS, WhatsApp59 (F)2026-11-14
6Braintrust AIRVoice59 (F)2026-09-25
7Fountain - AnnaVoice58 (F)2026-11-14
8Sense - AI Interviewer, AI Video InterviewerVoice, video57 (F)2026-11-14
9Sapia.ai - Chat InterviewText chat56 (F)2026-11-14
10Tenzo AI - MorganPhone, SMS, email, WhatsApp, video56 (F)2026-09-25
11Classet - JoyPhone55 (F)2026-09-25
12HackerRank - ChakraVoice54 (F)2026-11-14
13Glider AIVideo, voice52 (F)2026-09-25
14BrightHire - BrightHire ScreenVoice51 (F)2026-08-02
15Xobin - AI interviews, NURAInterview agent50 (F)2026-09-25
16Jobma - agentic AI interviewsVideo49 (F)2026-07-17
17CodeSignal - AI InterviewerVoice, text47 (F)2026-11-14
18Talview - IvyVideo, phone47 (F)2026-11-14
19Interviewer.AI - AI InterviewerVoice or video avatar47 (F)2026-11-14
20ConverzAI - Virtual RecruiterVoice, SMS, email44 (F)2026-09-25
21RibbonVoice, video43 (F)2026-09-25
22Humanly - AI InterviewerChat, voice, video42 (F)2026-11-14
23TengaiBrowser avatar42 (F)2026-09-25
24HirePro - AI RecruiterInterview agent38 (F)2026-09-25
25micro1 - ZaraVoice36 (F)2026-09-25
26HireLogic - AI interview agentAsync agent34 (F)2026-09-25
27Qureos - IrisVoice, video33 (F)2026-09-25
28Intervue.io - AI screeningVoice or avatar31 (F)2026-09-25

Phenom, Apriora and Braintrust AIR all round to 59, in that order on their unrounded scores. Sapia.ai and Tenzo AI both round to 56, as do CodeSignal, Talview and Interviewer.AI to 47, and Humanly and Tengai to 42; each group is listed in unrounded order.

Most AI interviewer bias audits never looked at a real applicant

Bias audits are where AI interviewer vendors look strongest on paper. Most of the top half score 68 or higher on bias audit transparency, and the dashboards show rows of "Clear" results. Read the reports and a pattern appears. Of the thirteen highest-scoring vendors:

  • Seven audit the interviewer on test or synthetic data. HeyMilo, Apriora, Braintrust AIR, Fountain, Tenzo AI and Glider AI publish Warden AI audits run on test data. The reports for HeyMilo, Apriora, Braintrust AIR and Tenzo AI say that was done "due to a lack of access to historical data"; Fountain's says historical data with enough demographic coverage wasn't available. Braintrust's earlier Holistic AI audit used 20,000 synthetic transcripts. Eightfold's BABL AI audit of its AI Interviewer uses historical data for gender, but synthetic LLM-generated interviews for race and ethnicity, which the auditor itself flags.
  • Four publish no current audit that covers the interviewer. Phenom's audits cover its Fit Score matching model, not the voice screening agent or AI Interviewer. Sense's Warden reports cover Candidate Matching and its Pre-Screening Bot, and name no audit for the AI Interviewer. HackerRank's only audits (BABL AI, July 2024) cover plagiarism detection and image analysis, not Chakra. Sapia.ai's only independent audit is from December 2022, before it moved to LLM scoring.
  • Two audit on real candidate data. Classet's Warden report for Joy uses historical data on real candidates, with demographics inferred from names. HireVue's audits of its AI-scored competencies use client data, but they are posted on customers' career sites rather than by HireVue, and they report impact ratios below 0.80 in some comparisons (57 of 640 in the 2025 DCI Consulting cycle).

Why it matters: a counterfactual test shows whether swapping a candidate's name or stated demographics changes the score. That is useful. It cannot show how real applicants from different groups fared once the tool met real accents, real connection quality and real job pools. One Warden report, Apriora's, says outright that it is not evidence of full NYC Local Law 144 compliance. Under LL 144 the audit, the published summary and the candidate notice remain the employer's duty, whoever the vendor hired.

1. Eightfold AI - 67/100

Eightfold AI is a talent-intelligence platform whose matching model is trained on more than a billion anonymised career profiles. It added an AI Interviewer (360 Interview) in 2026 alongside other agentic products.

Strengths. One of the best-documented vendors in the directory: ISO 42001 renewed in August 2026, a public status page and changelog, named EU AI Act documents including Instructions for Use backed by an independent assurance report, and bias audits for both the Matching Model and the AI Interviewer viewable without a request.

Watch out for. The AI Interviewer audit's race and ethnicity results rest on synthetic interviews. The Instructions for Use and Article 26 guidance sit behind a request form. Eightfold also faces Kistler v. Eightfold AI, a 2026 class action alleging that its match scores are undisclosed consumer reports under the FCRA.

2. HireVue - 62/100

HireVue is one of the most widely deployed AI hiring platforms in enterprise recruiting. Its voice-based AI Interviewer launched in June 2026, on top of its long-running AI-scored video interviews and assessments.

Strengths. The longest audit record in this list, on real client data, with annual audits from 2023 to 2026. It publishes a 29-page Explainability Statement, a per-product AI inventory, release notes that disclose model retrains, and a candidate opt-out from AI scoring.

Watch out for. HireVue does not post its LL 144 audit summaries itself; they live on employers' career sites, and the 2026 cycle came from a new auditor with an imprecise data window. Several technical reports sit in its trust centre rather than in public. Compare it directly with Apriora.

3. HeyMilo - 60/100

HeyMilo, a New York company founded in 2023, sells AI screening agents to high-volume teams: voice, phone and video interviews in about 20 languages, with per-question scores and the rationale behind them.

Strengths. A monthly Warden AI audit with a public dashboard, results for categories beyond LL 144's (age, disability, English proficiency, national origin), a public help centre and status page, and optional cheat detection.

Watch out for. The audit runs on test data and covers only the interview match score, not its resume scoring, speech score or cheat detection. Its video product page markets on-camera scoring of presence and eye contact that its help centre's transcript-based scoring doesn't describe, and its privacy policy says it collects no biometrics despite voice-biometric identity checks.

4. Phenom - 59/100

Phenom combines career sites, a candidate CRM and internal mobility with a growing set of AI agents, including a voice screening agent that scores answers from 1 to 5.

Strengths. Its Fit Score matching model has been audited every year since 2023, most recently by Idiro Analytics on more than 12 million 2025 applications, and its trust centre lists model cards for about 25 features.

Watch out for. None of that evidence covers the interviewers: there is no published audit for the voice screening agent or AI Interviewer, and nearly all the model cards sit behind a request form.

5. Apriora (Alex) - 59/100

Apriora, which trades as Alex, runs live two-way first-round interviews over video, phone, SMS and WhatsApp in more than 26 languages and says it has interviewed over a million candidates.

Strengths. Alex says interviews are scored on the transcript alone, with no facial or voice analysis. It publishes a docs site with a compliance FAQ, a dated subprocessor list, a status page and monthly Warden AI audits.

Watch out for. The audits run on test data and don't cover its Proctor cheating score, which tracks head and eye movement, or its Identity face matching. It promises no candidate is auto-rejected by AI, while its docs show score-based rejection rules.

6. Braintrust AIR - 59/100

Braintrust AIR runs adaptive voice interviews that are recorded, transcribed and scored 1 to 5 against recruiter-editable rubric criteria, then synced to more than 50 ATSs.

Strengths. It excludes emotion inference, facial recognition and voice-print matching, and publishes both a Holistic AI audit and a Warden AI summary without a request.

Watch out for. Its product page says "AIR does not auto-reject candidates", but a help-centre article tells recruiters it lets them "filter out unqualified candidates automatically". Both audits use synthetic or test records, and no published test covers how its language-proficiency, communication and AI-assistance signals affect candidates by national origin or speech-related disability. Its compliance page headlines "Zero bias detected" and "EEOC Compliant", and still cites a Colorado "SB 169" effective February 2026.

7. Fountain - 58/100

Fountain sells a high-volume hiring platform for frontline employers. Its voice agent Anna interviews applicants and scores them 1 to 5 against the employer's qualifications.

Strengths. Per-product Trust Cards that say the AI Recruiter is treated as high-risk, an announced ISO 42001 certificate, a public help centre, DPA and subprocessor list, and published Warden AI reports.

Watch out for. The audits use counterfactual test profiles rather than live applicants, and we found no second evaluation of the AI Recruiter after 1 June 2026. Anna scores English proficiency by default, which employers should weigh against national-origin discrimination rules.

8. Sense - 57/100

Sense is a talent engagement platform for high-volume and staffing recruitment that now sells an AI Interviewer and an AI Video Interviewer among its agents.

Strengths. A public help centre, status page and monthly release notes, plus October 2026 Warden AI reports.

Watch out for. Those reports cover Candidate Matching and the Pre-Screening Bot on test data; no audit is named for the AI Interviewer or AI Video Interviewer. Sense's own help documentation shows AI agents, when configured to, setting candidate dispositions without a recruiter step.

9-13. Sapia.ai, Tenzo AI, Classet, HackerRank, Glider AI

  • Sapia.ai - 56. A text-only Chat Interview with no video or audio, scored by an LLM that shows the excerpt behind each score; ISO 42001 certified. Its only independent audit predates its LLM scoring, so it is too old to support current LL 144 use.
  • Tenzo AI - 56. Morgan interviews by phone, text, email, WhatsApp or video in more than 40 languages. Its bias testing scores 83, second only to Eightfold in this set, though on test data; but its documentation is thin, its candidate terms still provide for employer-deployed "emotion recognition or biometric categorization features", and its Platform Agreement puts notices, consents and human review on the customer.
  • Classet - 55. Joy phones hourly applicants within about a minute and returns a qualified or unqualified outcome from employer-scripted knockout questions. It is the only vendor in the top half to publish its own audit on real candidate data, yet a company blog post argues Joy sits outside LL 144's AEDT definition.
  • HackerRank - 54. Chakra runs voice interviews for developer hiring, scores on a five-point scale and grades AI fluency, explaining each score with transcript evidence. No published audit covers Chakra.
  • Glider AI - 52. AI video and voice interviews with optional automatic disqualification below a cut-off score. A September 2026 Warden audit of the AI Interviewer exists, on test data, but its 2023 page saying "Glider is not an AEDT" is still online.

The rest of the field

BrightHire (51), an interview intelligence platform, added BrightHire Screen, a voice interviewer; it commissions BABL AI audits scoped to Screen, but keeps them behind a gated portal. Xobin (50) sells AI interviews from about $5 each and NURA, a recruiting agent launched in July 2026. Jobma (49) offers agentic AI interviews alongside one-way video. CodeSignal (47) sells AI Interviewer agents for voice or text, and Talview (47) its interviewer Ivy. Interviewer.AI (47) pairs a conversational avatar with one-way video scoring. ConverzAI (44) and Ribbon (43) sell voice agents to staffing and high-volume teams, and Humanly (42) can disqualify candidates on configurable score thresholds. Tengai (42) runs a 20-minute avatar interview whose "fully EU AI Act compliant" claim rests on an argument rather than documents. HirePro (38), micro1 (36), HireLogic (34), Qureos (33) and Intervue.io (31) score lowest, mostly on thin bias-audit and documentation evidence; HireLogic has no independent audit of its AI interview scores, and Qureos's interviews include Big Five personality scoring.

What AI interviewers score candidates on

Most tools in this list score the transcript against a rubric. Several go further, and that is where the risk sits:

  • Language and delivery. Fountain's Anna scores English proficiency by default; Braintrust AIR adds language-proficiency, communication and AI-assistance signals; HeyMilo has a separate speech score for clarity and delivery. Scores like these can disadvantage candidates by national origin or speech-related disability unless they are tested for it.
  • Integrity and identity. Apriora's Proctor tracks head and eye movement, Glider's proctoring checks eye movement, extra faces and voices, and several tools match a face to a government ID. In Illinois that brings in the state's biometric privacy law.
  • Emotion. The EU AI Act has banned emotion recognition at work since 2 February 2025. No vendor here says it scores emotion, but check the contract: Tenzo's candidate terms and Classet's "sentiment" API label both need an answer before any EU use.

The rules that apply to AI interviews

An AI interviewer that scores or ranks candidates is squarely in scope of the AI hiring laws:

  • NYC Local Law 144 treats a score used as a substantial factor in a hiring decision as an automated employment decision tool: an independent bias audit within the prior year, a published summary and candidate notice. A December 2025 New York State Comptroller audit found the city's enforcer had received only two complaints and, reviewing 32 companies, found one instance of non-compliance; the city agreed to adopt most of the audit's 13 recommendations.
  • Illinois has two laws in play: the AI Video Interview Act requires notice and consent before AI analyses a video interview, and HB 3773 has required notice of AI use in employment decisions since 1 January 2026.
  • California's Civil Rights Council rules on automated-decision systems have applied since 1 October 2025, with four-year record keeping.
  • Colorado's SB 26-189 adds notice, an explanation of adverse decisions and human review from 1 January 2027.
  • The EU AI Act treats interview scoring as high-risk under Annex III; those obligations apply from 2 December 2027.

How to choose an AI interviewer

  1. Ask what data the bias audit used. Test data, synthetic interviews and real applicant outcomes answer different questions. If the vendor has never audited on real outcomes, plan to run your own on your first months of data.
  2. Ask which scores the audit covers. An audit of the interview score says nothing about the speech score, the fraud flag or the resume match that sits next to it.
  3. Ask whether it can reject or advance anyone on its own. Several tools here can set dispositions or disqualify below a threshold when configured to. Decide where a person must sign off, and check that the product enforces it.
  4. Get the full list of what it scores. Transcript only, or also accent, fluency, delivery, eye movement and identity?
  5. Read who owns compliance in the contract. Notices, consents and impact assessments usually land on you. Ask for templates.

For the documentation questions, use our Annex IV walkthrough; for audit quality, see what makes an LL 144 audit defensible; and to check whether your use is high-risk under the EU AI Act, run the quick check. The wider market is ranked in the best AI recruiting tools.

If you're the candidate

AI interviewers are real and increasingly common, so being interviewed by one is not a red flag in itself. What you're owed depends on where you are. In New York City you should get notice at least 10 business days before an automated tool is used, with instructions for requesting an alternative process or an accommodation. In Illinois you should be told when AI is used, and asked for consent before AI analyses a video interview. From 2027, Colorado residents can ask for an explanation of an adverse decision and a human review.

If you're unsure whether you're talking to a person, ask. Ask, too, whether a person reviews the result and whether you can take a different route. Some vendors offer one: HireVue publishes a candidate opt-out from AI scoring.

Every score above links to a full scorecard with the evidence behind each category. Scores change when vendors publish new evidence, which any vendor can submit for review.

Frequently asked questions

What is an AI interviewer?

Software that conducts the interview itself - over the phone, on video with an avatar, or by text chat - asks the questions, often adapts its follow-ups, and returns a score, transcript and recording to the hiring team. It is different from a one-way video platform, where candidates record answers to fixed questions, and from an AI note-taker, which only records interviews that people run.

What is the best AI interviewer in 2026?

On HireAIScore's rubric, Eightfold AI scores highest at 67/100, followed by HireVue (62) and HeyMilo (60). The scores rate each vendor's public evidence - documentation, bias audits, deployer support, data governance, oversight and monitoring - not how well the interviewer performs, and none of the 28 tools reaches a C.

Are AI interviewer bias audits reliable?

Read them closely. Of the thirteen highest-scoring vendors, seven audit their interviewer on test or synthetic data rather than real applicants, and four publish no current audit covering the interviewer at all. A test-data audit shows whether changing a candidate's demographic details changes the score; it cannot show how real applicants from different groups actually fared.

Do employers have to tell you when an AI interviews you?

In several places, yes. NYC Local Law 144 requires notice at least 10 business days before an automated employment decision tool is used. Illinois HB 3773 requires notice when AI is used in employment decisions, and Illinois's AI Video Interview Act requires notice and consent before AI analyses a video interview. Colorado's SB 26-189 adds a point-of-interaction notice from 1 January 2027.

Is an AI interview a red flag?

Not by itself. AI interviewers are now sold by many mainstream hiring vendors. What matters is whether you were told, whether a person reviews the result before you are rejected, what the tool scores you on, and whether you can ask for an alternative process or an accommodation.

How were these AI interviewers ranked?

Each vendor is scored on HireAIScore's rubric v1.0 from public evidence: Article 11 technical documentation (20%), bias audit transparency (18%), FRIA support (15%), data governance disclosure (15%), human oversight design (12%), post-market monitoring (12%) and customer documentation (8%). The ranking is the overall score, which covers the vendor's AI hiring features as a whole, not the interviewer alone.