# Eightfold AI: HireAIScore fact sheet

URL: https://www.hireaiscore.com/vendors/eightfold-ai
Fact sheet: https://www.hireaiscore.com/vendors/eightfold-ai/facts.md
Last reviewed: 2026-11-14
Rubric: v1.0 (https://www.hireaiscore.com/methodology)

> HireAIScore rates Eightfold AI 67 out of 100 (grade D, Concerning) under rubric v1.0, last reviewed November 14, 2026. Eightfold AI ranks 1 of 86 vendors rated in Candidate sourcing AI, against a category average of 46. That score is drawn from 35 evidence items across 7 rubric criteria - 29 with a cited source, 6 recording that nothing was located.

This file is the text of the profile at the URL above, for AI assistants and other software. Scores are an editorial assessment of public evidence against a published rubric. They are not legal advice, a certification, or a statement that a product is or is not compliant.

## Score

- Overall: 67/100, grade D (Concerning)
- Rank: 1 of 86 in Candidate sourcing AI (category average 46)

## Criterion scores and evidence

### Article 11 Technical Documentation: 76/100 (weight 20%)

- Documentation (captured 2026-11-14): Wolfia trust portal (launched April 2026) lists EU AI Act Instructions for Use, an Independent Limited Assurance Report on Instructions for Use, EU AI Act Article 26 Guidance, EU AI Act Article 50 Transparency Obligations, Responsible AI Practices, Responsible AI for AI Interviewer and the ISO 42001 certification audit report, all behind Request access. Source: https://trust.eightfold.ai/documentation
- Public statement (captured 2026-11-14): Trust portal update of 22 August 2026: ISO/IEC 27001, 27701 and 42001 certifications renewed for 2026. Source: https://trust.eightfold.ai/updates
- Public statement (captured 2026-11-14): ISO/IEC 42001:2023 certification across developer, provider and consumer roles (August 2025); certification body not named. Source: https://eightfold.ai/company/press/press-releases/eightfold-ai-achieves-all-three-levels-of-iso-iec-420012023-certification/
- Audit report (captured 2026-11-14): The BABL AI audit of AI Interviewer publishes a system description: an LLM generates questions from the job description and resume, speech is transcribed, and a second LLM scores the transcript against an LLM-generated, human-editable 0 to 5 rubric; hiring managers see only the overall score. Source: https://eightfold.ai/wp-content/uploads/eightfold-ai-interviewer-summary-of-bias-audit-results.pdf
- Searched, not found (captured 2026-11-14): No public Annex IV-aligned technical file, model card or system card for the Matching Model or AI Interviewer. Searched: eightfold.ai/trust, eightfold.ai/trust/responsible-ai, trust.eightfold.ai/documentation, web search for Eightfold model card, AI fact sheet or system card.

### Bias Audit Transparency: 86/100 (weight 18%)

- Audit report (captured 2026-11-14): BABL AI NYC LL 144 audit of the Eightfold Matching Model (v1.0 3 March 2026): data January 2024 to December 2025, about 23.8 million applicants with known gender and 15.5 million with known race, intersectional tables, lowest impact ratio 0.880 (Asian female). Source: https://eightfold.ai/wp-content/uploads/eightfold-summary-of-bias-audit-results.pdf
- Audit report (captured 2026-11-14): BABL AI NYC LL 144 audit of AI Interviewer (opinion as of 29 June 2026, signed 8 July 2026): gender results use historical self-reported data; race/ethnicity and intersectional results use synthetic LLM-generated interviews, which the auditor flags as an Emphasis of Matter. All impact ratios are at or above 0.8. Source: https://eightfold.ai/wp-content/uploads/eightfold-ai-interviewer-summary-of-bias-audit-results.pdf
- Audit report (captured 2026-11-14): Trust portal makes three bias audit reports publicly viewable without a request: EF Matching Model, AI Interviewer and Extended Categories; the 30 August 2026 update states annual audits under NYC LL 144. Source: https://trust.eightfold.ai/documentation
- Audit report (captured 2026-11-14): Customer-side NYC LL 144 audit of Morgan Stanley's use of the Eightfold model by BLDS, LLC (July 2026): applicant data 1 June 2025 to 31 May 2026, impact ratios from 1.000 down to 0.864. Source: https://www.morganstanley.com/content/dam/msdotcom/en/disclosures/Bias_Audit_of_Morgan_Stanleys_Use_of_Eightfold_Model_For_Scoring_Applicants.pdf
- Audit report (captured 2026-11-14): Trust page summarising the 2026 Matching Model audit, with a link to the AI Interviewer results. Source: https://eightfold.ai/trust/bias-audit-results/

### FRIA Support: 45/100 (weight 15%)

- Documentation (captured 2026-11-14): Gated deployer documents named in the trust portal: EU AI Act Instructions for Use, EU AI Act Article 26 Guidance, EU AI Act Article 50 Transparency Obligations, Data Protection Impact Assessment Report and a Transfer Impact Assessment Reference Guide for Customers. Source: https://trust.eightfold.ai/documentation
- Documentation (captured 2026-11-14): January 2026 deployer-obligations post with shared-responsibility framing; it still describes the move of Annex III obligations to December 2027 as a proposal awaiting approval, which the Digital Omnibus on AI (Regulation (EU) 2026/1744) has since adopted. Source: https://eightfold.ai/blog/eu-ai-act-hr-leaders/
- Documentation (captured 2026-11-14): July 2026 release: always-on AI disclosure labels across Eightfold's assistants, agents and AI Interviewer, built for Article 50 and not editable by customers. Source: https://releases.eightfold.ai/announcements/ai-transparency-labels-across-eightfold-products
- Searched, not found (captured 2026-11-14): No FRIA or Article 27 template or FRIA-mapping guidance. Searched: trust.eightfold.ai/documentation (no FRIA item), eightfold.ai/trust/responsible-ai, eightfold.ai/blog/eu-ai-act-hr-leaders, web search for Eightfold FRIA deployer guidance.

### Data Governance Disclosure: 65/100 (weight 15%)

- Public statement (captured 2026-11-14): States the platform does not use identifiable customer data to train its models; models improve on anonymised, aggregated data. Source: https://eightfold.ai/trust/security/
- Public statement (captured 2026-11-14): States the matching model is trained on more than one billion anonymised career profiles and that bias is tested on each model update. Source: https://eightfold.ai/trust/responsible-ai/
- Documentation (captured 2026-11-14): Public help article: resume masking removes names, gender, ethnic background, age, education, emails, links and pronouns, and is no longer a configurable feature. Source: https://docs.eightfold.ai/enhanced-resume-masking-documentation
- Documentation (captured 2026-11-14): June 2026 release adds a retention rule that deletes AI interview recordings and transcripts on a set schedule, independently of the candidate profile. Source: https://releases.eightfold.ai/announcements/data-retention-and-purge-for-ai-interviewer
- Documentation (captured 2026-11-14): Gated: Candidate Data Anonymization for Model Training, Data Anonymization Report, Matching Model Outcomes (January 2025 and September 2026) and the DPIA report. Source: https://trust.eightfold.ai/documentation
- Searched, not found (captured 2026-11-14): No public breakdown of training-data sources, no validation accuracy by subgroup and no fixed retention periods for customer data. Searched: eightfold.ai/trust/responsible-ai, eightfold.ai/trust/security, eightfold.ai/privacy-policy (May 2026), Eightfold responsible-AI blog posts from 2026.

### Human Oversight Design: 64/100 (weight 12%)

- Documentation (captured 2026-11-14): Public help article: each match score comes with an explanation of up to ten hiring criteria with positive, negative or neutral signals; insights are directional and should complement, not replace, recruiter judgment. Source: https://docs.eightfold.ai/match-score-explanation
- Documentation (captured 2026-11-14): Recruiters or hiring managers calibrate a position (ideal candidates, skills, titles, experience), which drives the match score; an automatic calibration mode needs no user input. Source: https://docs.eightfold.ai/position-calibration-overview
- Documentation (captured 2026-11-14): AI Interviewer feedback, transcript and recording are reviewed by a recruiter in the platform. Source: https://docs.eightfold.ai/en_US/ai-interviewer/ai-interviewer-overview
- Public statement (captured 2026-11-14): States the software does not autonomously reject a candidate or advance anyone through a hiring stage on its own. Source: https://eightfold.ai/trust/responsible-ai/
- Searched, not found (captured 2026-11-14): No public documentation of overriding a match or interview score, configurable confidence thresholds, or a per-tenant switch to turn matching off. Searched: docs.eightfold.ai match-score, calibration and AI Interviewer articles and site search; some admin articles redirect to customer sign-in.

### Post-Market Monitoring: 58/100 (weight 12%)

- Documentation (captured 2026-11-14): Public status page with regional components, incident history and email, Slack and RSS subscriptions (status.eightfold.ai does not resolve). Source: https://eightfold.statuspage.io/
- Documentation (captured 2026-11-14): Public product changelog back to March 2025, each change tagged default on, default off, non-deferrable or deprecation, with email subscription; it covers features, not model versions. Source: https://releases.eightfold.ai/
- Documentation (captured 2026-11-14): Security terms commit to notifying customers of security incidents with a goal of 24 hours and no later than 72 hours from confirmation. Source: https://eightfold.ai/wp-content/uploads/security-terms.pdf
- Documentation (captured 2026-11-14): Subscribable trust-portal feed announcing new audits, certifications and outcome reports (for example Matching Model Outcomes, September 2026). Source: https://trust.eightfold.ai/updates
- Public statement (captured 2026-11-14): ISO 42001 scope includes monitoring and incident response for AI behaviour. Source: https://eightfold.ai/company/press/press-releases/eightfold-ai-achieves-all-three-levels-of-iso-iec-420012023-certification/
- Searched, not found (captured 2026-11-14): No public AI-incident or AI-harm reporting channel separate from security incidents, and no versioned changelog for the Matching Model. Searched: eightfold.ai/trust, eightfold.ai/trust/security, eightfold.ai/.well-known/security.txt (404), releases.eightfold.ai June to October 2026.

### Customer Documentation: 66/100 (weight 8%)

- Documentation (captured 2026-11-14): Public help centre and documentation portal (some articles require customer sign-in) linking release notes, community and API documentation. Source: https://docs.eightfold.ai/
- Documentation (captured 2026-11-14): Intended-use statement for AI Interviewer, which primarily supports profile screening and custom questions. Source: https://docs.eightfold.ai/en_US/ai-interviewer/ai-interviewer-overview
- Documentation (captured 2026-11-14): NYC LL 144 deployer page for the matching model: inputs, explanations, a retention pointer and a statement that the model does not replace human decision making. Source: https://eightfold.ai/nyc-eightfold-matching-model/
- Searched, not found (captured 2026-11-14): No public Article 13 instructions for use and no consolidated limitations statement covering the Matching Model across external hiring, internal mobility and career-site personalisation. Searched: trust.eightfold.ai/documentation (gated), docs.eightfold.ai search, eightfold.ai/trust/responsible-ai.

## Company facts

- Website: https://eightfold.ai
- Founded: 2016
- Headquarters: United States (US)
- Pricing tier: Enterprise
- Market side: Employer-side (AEDT)
- Categories: Candidate sourcing AI, ATS-integrated AI
- Description: Talent intelligence platform offering candidate matching, internal mobility, and career-site personalization on top of a deep-learning talent graph.

These are company details, not findings, and carry no weight in the score.

## Head-to-head comparisons

- [Eightfold AI vs Greenhouse](https://www.hireaiscore.com/compare/eightfold-ai-vs-greenhouse): 67 vs 67
- [Eightfold AI vs Beamery](https://www.hireaiscore.com/compare/beamery-vs-eightfold-ai): 67 vs 66
- [Eightfold AI vs SmartRecruiters](https://www.hireaiscore.com/compare/eightfold-ai-vs-smartrecruiters): 67 vs 62
- [Eightfold AI vs Phenom](https://www.hireaiscore.com/compare/eightfold-ai-vs-phenom): 67 vs 59
- [Eightfold AI vs Workday Recruiting AI](https://www.hireaiscore.com/compare/eightfold-ai-vs-workday-recruiting-ai): 67 vs 59
- [Eightfold AI vs HiredScore](https://www.hireaiscore.com/compare/eightfold-ai-vs-hiredscore): 67 vs 58

## Conflicts of interest

No vendor pays for placement, scoring, or removal. Casework - the consulting firm that operates this directory - provides paid services to some vendors. Any active or recent (within 24 months) commercial relationship is disclosed on the affected vendor profile and the review is reassigned to an independent reviewer.

Casework has no commercial relationship with this vendor.

Full policy: https://www.hireaiscore.com/about

## How to cite

HireAIScore, "Eightfold AI" profile, rubric v1.0, last reviewed 2026-11-14. https://www.hireaiscore.com/vendors/eightfold-ai

Please link the profile URL rather than this file, quote the score with its rubric version and review date, and include the vendor's reply where one is published.
