Video interviews scored by AI
What audits, tests and legal cases found, and how notice, consent and audit rules work now
IAPQB Report No. 9 · Practice report · Published 17 November 2026
International AI Practice Qualification Board (IAPQB)
In one minute
- Some hiring tools score recorded video answers. A newsroom test found scores that moved with glasses, a headscarf or a bookshelf [1].
- HireVue dropped visual analysis from new assessments in 2020; the audit it released in 2021 covered, by its own account, one use case [2, 3].
- Courts in Massachusetts are testing whether AI interviews can be unlawful lie detector tests. HireVue and Intuit dispute a Colorado complaint about a deaf employee's interview [4, 5, 6].
- Illinois requires notice and consent, Maryland a signed waiver for facial recognition, New York City notice and yearly bias audits, and the EU bans emotion recognition in recruitment. Illinois received no demographic data under its law for 2024 or 2025 [7, 8, 9, 10, 11].
1. A familiar situation
A graduate applies for a trainee job. The invitation asks her to record answers to five questions on her laptop. She records at her kitchen table, with a lamp behind her. A week later an email rejects her. She does not know whether anyone watched the video, whether software scored her words, voice or face, or whom to ask.
2. Why it happens
A video interview tool turns words, voice or picture into scores, learning from past examples. Retorio, a German start-up, said its system learned from videos of more than 12,000 people, rated by 2,500 others for the impression they made [1]. Such a model can pick up cues that swayed those raters [1]. Speech recognition may also work less well for some groups: the Colorado complaint cites studies finding more errors for deaf and non-white speakers [12].
The UK Information Commissioner's Office (ICO) audited a provider whose tool scored written interview answers. Its privacy notice and its clients' notices pointed at each other, so neither informed candidates properly [13].
IAPQB view. In each case below, details of what a video system measured emerged only after it was in use, through an audit, a newsroom test, lawsuits or a complaint [1, 3, 4, 12]. In Colorado the parties disagreed about whether AI assessed the interview at all [6, 12].
3. Four cases
3.1 United States, 2019 to 2021: HireVue drops face analysis and publishes a narrow audit
In November 2019 the Electronic Privacy Information Center (EPIC) complained to the US Federal Trade Commission about HireVue, saying the company denied using facial recognition while analysing facial expressions [14]. In January 2021 HireVue said it had dropped visual analysis from new assessments early in 2020. It released an audit by O'Neil Risk Consulting and Algorithmic Auditing (ORCAA) and called the audited assessment "a representative use case" [2]. The audit says it "was not a comprehensive audit" and that the audited case was "not necessarily common or representative" [3]. Very short answers, more common among minority candidates, had been flagged for human review without a score; HireVue then had the algorithm score them too. ORCAA did not check the fixes itself [3].
3.2 Germany, 2021: glasses, a headscarf and a bookshelf change the scores
Bayerischer Rundfunk (BR), a German public broadcaster, tested a free trial of Retorio's personality-scoring software in late 2020 and reported in February 2021 [1]. An actress gave the same answers in each video. Glasses cut her conscientiousness score by nearly 10 points; a headscarf moved her scores by nearly 20 points. Repeats without changes gave stable results. A picture or bookshelf in the background, or a darker image, also changed other people's scores [1, 15]. Retorio said such factors also count in a normal interview [1].
3.3 Massachusetts, 2024 to 2026: a lie detector law meets AI interviews
Massachusetts bans lie detector tests in hiring and requires a notice saying so on job applications [4]. A CVS applicant sued, alleging that his 2021 HireVue interview went to Affectiva, which "analyzes candidates' facial expressions, eye contact, voice intonation, and inflection" [4]. Judge Patti B. Saris let the case continue on 16 February 2024; CVS had not challenged the lie detector claims [4, 16]. The applicant settled in July 2024, on terms not made public [16, 17]. A similar claim against JPMorgan Chase was dismissed on 22 October 2025 because the allegations showed only that HireVue interviews "could constitute lie detector tests" if used for that purpose [5]. The appeal was heard on 28 July 2026 [18].
3.4 Colorado, 2025: a deaf employee asks for captions
On 19 March 2025 the American Civil Liberties Union (ACLU) of Colorado filed a complaint with state and federal agencies for a Deaf, Indigenous Intuit employee [6]. It says she asked for human captions for a 2024 HireVue interview for a promotion, that the subtitles Intuit offered were not available, and that after rejection she got feedback, which she believed was automated, to "[p]ractice active listening". It alleges that AI scored the interview using speech recognition [12]. HireVue's chief executive said "Intuit did not use a HireVue AI-based assessment"; Intuit said the claims had no merit [6].
4. How practitioners handle it now
Notice and consent come before the interview. Since 1 January 2020 Illinois has required employers that use AI to analyse video interviews for Illinois jobs to tell applicants in advance, explain how the AI works and "what general types of characteristics it uses", and get consent [7, sec. 5]. Maryland requires a signed waiver before facial recognition is used to create a facial template during an interview [8]. New York City requires notice 10 business days before an automated hiring tool is used [19].
Candidates are told how to ask for another route. New York City's notice must explain how to request "an alternative selection process or a reasonable accommodation under other laws, if available", though employers need not offer an alternative [9]. After the audit, HireVue planned to point candidates to the employer's diversity and inclusion team to discuss options [3].
Face and emotion reading are restricted or dropped. Since 2 February 2025 the EU has banned AI that infers emotions at work from biometric data such as faces or voices, except for medical or safety reasons. The European Commission's non-binding guidelines say: "Using emotion recognition AI systems during the recruitment process is prohibited" [20, art. 5; 10, paras 5, 250, 251, 254 and 429]. Separate high-risk rules for recruitment AI apply from 2 December 2027 under the Digital Omnibus on AI [20; 21].
Bias audits are independent and published, and enforcement is checked. Since July 2023 New York City has required a yearly independent bias audit of automated hiring tools, with selection or scoring rates and impact ratios by sex and by race or ethnicity, and a published summary [9, 19]. In December 2025 the New York State Comptroller reported two complaints in two years; reviewing 32 companies, the city found one problem and state auditors at least 17 possible ones. The city closed one complaint on the employer's sworn word that its AI-marketed tool only recorded interviews for human review, without examining the software [22].
Sharing is limited, videos are deleted on request, and outcomes must be reported. Illinois limits who may receive the videos and requires deletion within 30 days of a request, including copies held by others [7, secs. 10 and 15]. Since 2022, employers that rely only on AI to choose who gets an in-person interview must report applicants' race and ethnicity each year [7, sec. 20]. The state's reports for 2024 and 2025 say no such data was reported, and give no reason [11]. The Act sets no penalty [7].
5. What to take from this
For recruiters and hiring managers. Ask the vendor what the tool measures in the employer's set-up and whether a person reviews each result, and say so in the invitation [7, sec. 5]. In Colorado and New York City, that answer came after a complaint [6, 22].
For HR leaders who buy these tools. Read the full audit report and its caveats; HireVue's release did not repeat them [2, 3]. Name in the contract who informs candidates and who handles requests for another route [9, 13].
For careers services and training providers. Teach candidates to ask whether AI will assess the interview, how to request an alternative and, in Illinois, how to have the video deleted [9; 7, sec. 15].
6. Limits of this report
Three cases are American and one German; rules elsewhere differ. Lawsuits and complaints contain allegations; some facts come from news reports and the parties' statements. The CVS case settled without findings [5], and the JPMorgan appeal was undecided in the last report found [18]. The BR test was small and made no general claim [1]. The ORCAA audit dates from 2020 [3]. This report does not test any AI tool.
References
[1] Harlan, E., & Schnuck, O. (16 February 2021). Objective or Biased: On the questionable use of Artificial Intelligence for job applications. BR Data, Bayerischer Rundfunk, a joint investigation with report München. https://interaktiv.br.de/ki-bewerbung/en/
[2] HireVue (12 January 2021). Hirevue leads the industry with commitment to transparent and ethical use of AI in hiring [press release, Salt Lake City]. https://www.hirevue.com/press-release/hirevue-leads-the-industry-with-commitment-to-transparent-and-ethical-use-of-ai-in-hiring
[3] O'Neil Risk Consulting and Algorithmic Auditing (ORCAA) (15 December 2020). Description of Algorithmic Audit: Pre-built Assessments. Published by HireVue. https://www.hirevue.com/wp-content/uploads/2021/01/oneil-risk-consulting-and-algorithmic-auditing-01-2021.pdf
[4] Baker v. CVS Health Corporation and CVS Pharmacy, Inc., Civil Action No. 23-11483 (D. Mass.), decision of 16 February 2024 (Saris, J.) [page title: BAKER v. CVS HEALTH CORPORATION CVS (2024)]. FindLaw. https://caselaw.findlaw.com/court/us-dis-crt-d-mas/115835556.html
[5] Saint Cyr v. JPMorgan Chase Bank, N.A., No. 1:25-cv-11751-BEM (D. Mass.), Memorandum and Order on Defendant's Motion to Dismiss, 22 October 2025. Read in the addendum to the Brief for Plaintiff-Appellant, Saint Cyr v. JPMorgan Chase & Co., No. 25-2129 (1st Cir.), filed 5 February 2026, as published by Courthouse News Service. https://www.courthousenews.com/wp-content/uploads/2026/07/hirevue-first-circuit-class-action-appellant.pdf
[6] Kalser, L. (24 March 2025). AI hiring software was biased against deaf employees, ACLU alleges in ADA case. HR Dive. https://www.hrdive.com/news/ai-intuit-hirevue-deaf-indigenous-employee-discrimination-aclu/743273/
[7] Illinois General Assembly. (820 ILCS 42/) Artificial Intelligence Video Interview Act [current text of sections 1, 5, 10, 15 and 20, with source notes P.A. 101-260, eff. 1-1-20, and P.A. 102-47, eff. 1-1-22]. https://www.ilga.gov/Legislation/ILCS/Articles?ActID=4015&ChapterID=68 (accessed 16 November 2026)
[8] Maryland General Assembly (8 May 2020). Chapter 446 (House Bill 1202): Labor and Employment – Use of Facial Recognition Services – Prohibition. 2020 Laws of Maryland, effective 1 October 2020. https://mgaleg.maryland.gov/2020RS/Chapters_noln/CH_446_hb1202t.pdf. Current text: Maryland General Assembly. Laws - Statute Text [Article - Labor and Employment, section 3-717]. https://mgaleg.maryland.gov/mgawebsite/Laws/StatuteText?article=gle§ion=3-717&enactments=false (accessed 16 November 2026)
[9] New York City Department of Consumer and Worker Protection (2023). Notice of Adoption of Final Rule [Rules of the City of New York, Title 6, Chapter 5, Subchapter T: Automated Employment Decision Tools, sections 5-300 to 5-304; rule effective 5 July 2023]. https://rules.cityofnewyork.us/wp-content/uploads/2023/04/DCWP-NOA-for-Use-of-Automated-Employment-Decisionmaking-Tools-2.pdf (rule page: https://rules.cityofnewyork.us/rule/automated-employment-decision-tools-updated/)
[10] European Commission (29 July 2025). Commission Guidelines on prohibited artificial intelligence practices established by Regulation (EU) 2024/1689 (AI Act). C(2025) 5052 final. https://ec.europa.eu/newsroom/dae/redirection/document/112367, listed on the page "Commission publishes the Guidelines on prohibited artificial intelligence (AI) practices, as defined by the AI Act." https://digital-strategy.ec.europa.eu/en/library/commission-publishes-guidelines-prohibited-artificial-intelligence-ai-practices-defined-ai-act
[11] Illinois Department of Commerce and Economic Opportunity (July 2025). 2024 Artificial Intelligence Video Interview Act Report. Report to the 104th General Assembly. https://dceo.illinois.gov/content/dam/soi/en/web/dceo/aboutdceo/reportsrequiredbystatute/2024-ai-video-report.pdf. And (June 2026). 2026 Artificial Intelligence Video Interview Act Report. Report to the 104th General Assembly. https://dceo.illinois.gov/content/dam/soi/en/web/dceo/aboutdceo/reportsrequiredbystatute/2025-ai-video-report.pdf (the June 2026 report covers the 12 months to 30 November 2025 and is published under a file name ending in 2025)
[12] ACLU of Colorado, for the charging party (March 2025). Complaint of Discrimination [against Intuit, Inc. and HireVue, Inc., before the Colorado Civil Rights Division; redacted copy]. American Civil Liberties Union. https://assets.aclu.org/live/uploads/2025/03/Redacted-HireVue_Intuit-Complaint-of-Discrimination_Redacted.pdf
[13] Information Commissioner's Office (November 2024). AI tools in recruitment: Audit outcomes report. https://ico.org.uk/media2/migrated/4031620/ai-in-recruitment-outcomes-report.pdf
[14] Electronic Privacy Information Center (EPIC). In re HireVue [case page; complaint filed with the FTC on 6 November 2019]. https://epic.org/documents/in-re-hirevue/ (accessed 16 November 2026)
[15] Reuter, M. (19 February 2021). Algorithmen für Bewerbungen: Das Bücherregal im Hintergrund hat Einfluss darauf, ob Du den Job bekommst [in German]. netzpolitik.org. https://netzpolitik.org/2021/algorithmen-fuer-bewerbungen-das-buecherregal-im-hintergrund-hat-einfluss-darauf-ob-du-den-job-bekommst/
[16] Umanah, U. (18 July 2024). CVS Applicant Agrees to Settle AI-Aided Lie Detector Test Suit. Bloomberg Law. https://news.bloomberglaw.com/litigation/cvs-applicant-agrees-to-settle-ai-aided-lie-detector-test-suit
[17] Crist, C. (24 July 2024). CVS settles lawsuit alleging it used AI 'lie detector' in interviews. HR Dive. https://www.hrdive.com/news/cvs-settles-lawsuit-over-using-ai-based-lie-detector/722249/
[18] Russell, J. (28 July 2026). Job seeker asks First Circuit to revive class action over AI interviewing tool. Courthouse News Service. https://courthousenews.com/job-seeker-asks-first-circuit-to-revive-class-action-over-ai-interviewing-tool/
[19] New York City Department of Consumer and Worker Protection (29 June 2023). Automated Employment Decision Tools: Frequently Asked Questions. https://www.nyc.gov/assets/dca/downloads/pdf/about/DCWP-AEDT-FAQ.pdf
[20] European Commission, AI Act Service Desk. Regulation (EU) 2024/1689 (AI Act); the article pages state that their text is based on the EUR-Lex consolidated version as at 27 July 2026 and mark the changes made by the Digital Omnibus on AI. Article 5: Prohibited AI practices: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-5. ANNEX III: https://ai-act-service-desk.ec.europa.eu/en/ai-act/annex-3. Article 113: Entry into force and application: https://ai-act-service-desk.ec.europa.eu/en/ai-act/article-113 (all accessed 16 November 2026)
[21] European Parliament and Council of the European Union (8 July 2026). Regulation (EU) 2026/1744 of the European Parliament and of the Council of 8 July 2026 amending Regulations (EU) 2024/1689, (EU) 2018/1139 and (EU) 2023/1230 as regards the simplification of the implementation of harmonised rules on artificial intelligence (Digital Omnibus on AI) (Text with EEA relevance). EUR-Lex. https://eur-lex.europa.eu/eli/reg/2026/1744/oj/eng
[22] Office of the New York State Comptroller (2 December 2025). Enforcement of Local Law 144 – Automated Employment Decision Tools. Report 2024-N-6. https://www.osc.ny.gov/state-agencies/audits/2025/12/02/enforcement-local-law-144-automated-employment-decision-tools (full report: https://www.osc.ny.gov/files/state-agencies/audits/pdf/sga-2026-24n6.pdf)
How to cite
IAPQB (2026). Video interviews scored by AI: what audits, tests and legal cases found, and how notice, consent and audit rules work now. IAPQB Report No. 9. International AI Practice Qualification Board. https://iapqb.org/research/009-ai-scored-video-interviews.html
About this report
Prepared by IAPQB. Each fact comes from the sources listed, and each source was opened and checked on 16 November 2026; anything that could not be checked was left out. AI tools were used to search and to draft, and IAPQB checked every claim against its source and is responsible for the text. IAPQB runs an AI work qualification; this report does not promote it. Corrections: contact@iapqb.org or an issue at github.com/iapqb/iapqb.github.io. Licence: CC BY 4.0.
Archived with the series on Zenodo: doi.org/10.5281/zenodo.24248199 · Markdown sources: github.com/iapqb/research