Xobin
HireAIScore rates Xobin 50 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Xobin ranks 21 of 82 vendors rated in Assessment AI, against a category average of 46. That score is drawn from 32 evidence items across 7 rubric criteria for Xobin — 27 with a cited source, 5 recording that nothing was located.
Score breakdown
Xobin, criterion by criterion
Each rubric criterion is scored out of 100 on the cited evidence, then weighted into the total.
Weighted contribution shown to the right of each bar.
Article 11 Technical Documentation
Weight 20% · category avg 50
62
+12.4
Bias Audit Transparency
Weight 18% · category avg 41
43
+7.7
FRIA Support
Weight 15% · category avg 31
35
+5.3
Data Governance Disclosure
Weight 15% · category avg 53
57
+8.5
Human Oversight Design
Weight 12% · category avg 55
58
+7.0
Post-Market Monitoring
Weight 12% · category avg 38
38
+4.6
Customer Documentation
Weight 8% · category avg 58
60
+4.8
Total weighted score
50.3(rounded to 50 for display)
Category avg is the mean raw score on that criterion across the 82 Assessment AI vendors in scope of this rubric, this one included.
Strongest criterion
Article 11 Technical Documentation
62raw / 100
Contributes 12.4 to the total. 62 against a 50 category average.
Weakest criterion
FRIA Support
35raw / 100
Contributes 5.3 to the total. 35 against a 31 category average.
Cited evidence
The evidence behind the score
Every score is backed by at least one cited piece of evidence. Open a criterion to see what was found, when, and where.
Download diligence recordEvidence ledger
- Items
- 32
- Documentation
- 23
- Public statement
- 4
- Absence
- 5
- With a source URL
- 27 of 32
- Source hostnames
- 3
- Fewest items
- 2
- FRIA Support
These figures measure how thoroughly Xobin was reviewed, not how Xobin performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
5 items62
- DocumentationCaptured Sep 25, 2026
Xobin's public AI Governance and Responsible AI Usage Policy (v1.2) documents governance roles, STRIDE threat modelling, DPIAs, a gateway-redaction-OpenAI inference architecture, drift monitoring, human-in-the-loop overrides and an audit trail of inputs, outputs, confidence and rationale.
- DocumentationCaptured Sep 25, 2026
The same policy says an EU AI Act technical dossier covering system details, intended use, logging, a post-market monitoring plan and OpenAI model cards will be provided only 'as and when requested', so it is not public.
- Public statementCaptured Sep 25, 2026
Xobin announced ISO 42001 certification on 3 September 2026, but the post names no certification body, certificate number, scope or validity dates and refers customers to support@xobin.com for a scope summary.
- DocumentationCaptured Sep 25, 2026
A public technical report on the Xobin Psychometric Test gives internal-consistency alphas of 0.75-0.88 (N about 5,000), six-month test-retest r = 0.81 (N = 120) and criterion correlations of about 0.20-0.35. The founder wrote it, and it covers the SJT rather than the AI scoring features.
https://xobin.com/research/articles/psychometric-test-validity-2025
- AbsenceCaptured Sep 25, 2026
No system or model card, instructions-for-use or public EU AI Act technical documentation for AI Evaluate, the AI interviewer or NURA was found on the AI policy, research hub, product pages, trust whitepaper or help center.
Bias Audit Transparency
4 items43
- DocumentationCaptured Sep 25, 2026
The AI policy publishes an internal fairness table for 1,020 interview sessions, labelled an 'illustrative validation cohort', with disparate impact ratios of 0.95-1.02 by gender, age band and English-language status. It gives no race/ethnicity breakdown, no dates and no external auditor.
- DocumentationCaptured Sep 25, 2026
The founder-authored psychometric test report says selection rates by gender and race have stayed within the four-fifths rule in large deployments, but it publishes no impact-ratio tables and names no auditor.
https://xobin.com/research/articles/psychometric-test-validity-2025
- DocumentationCaptured Sep 25, 2026
Xobin Research, which describes itself as an in-house team, says it monitors adverse-impact ratios and audits AI interview scoring for demographic drift. It publishes methodology, not audit results.
- AbsenceCaptured Sep 25, 2026
No NYC Local Law 144 bias-audit summary or independent third-party audit was found on xobin.com, its research hub, AI policy, trust whitepaper or help center, and trust.warden-ai.com/xobin returns 404.
FRIA Support
2 items35
- DocumentationCaptured Sep 25, 2026
The AI policy acknowledges high-risk classification under EU AI Act Annex III (employment) and describes a vendor DPIA and risk register covering risks to fundamental rights, plus a technical dossier available on request.
- AbsenceCaptured Sep 25, 2026
No EU AI Act Article 27 Fundamental Rights Impact Assessment template or deployer-obligation guidance was found in the AI policy, trust whitepaper, GDPR page, research hub or help center.
Data Governance Disclosure
5 items57
- DocumentationCaptured Sep 25, 2026
The AI policy says the AI is inference-only on Google Cloud using OpenAI models in zero-retention mode, redacts PII before model calls, never uses customer data for training, keeps data in customer-selected India, EU or US regions and retains assessments for 12 months by default.
- DocumentationCaptured Sep 25, 2026
The certifications page lists ISO 42001, SOC 2 Type 1 and Type 2, ISO 27001, GDPR, CSA STAR and six-monthly penetration testing, but gives no certificate numbers, auditors or downloadable reports.
- DocumentationCaptured Sep 25, 2026
The CSA STAR registry has listed Xobin since 2 July 2024 with a CAIQ submission, which is a Level 1 self-assessment rather than third-party STAR certification.
- DocumentationCaptured Sep 25, 2026
The privacy policy (updated 17 August 2026) sets an 8-year standard retention period for information Xobin controls and deletes customer service data within about nine months of account termination, but names no subprocessors.
- DocumentationCaptured Sep 25, 2026
A help-center article updated May 2026 says Xobin follows a 7-year data retention policy unless the customer asks for purging, with SLA-defined retention of 3 months to 7 years, which contradicts the AI policy's 12-month default.
Human Oversight Design
6 items58
- DocumentationCaptured Sep 25, 2026
The AI policy requires human-in-the-loop review: recruiters confirm, re-run or override AI draft scores, which come with confidence bands and rationale, using reason codes, and overrides are logged and sampled in monthly audits.
- DocumentationCaptured Sep 25, 2026
Controlled AI interviews score each answer against the customer's rubric with evidence quoted from the transcript, and the page says a human can override any score.
- DocumentationCaptured Sep 25, 2026
Guided AI interviews use a compliance filter to block protected-characteristic questions, keep every transcript reviewable and let candidates request a human round.
- DocumentationCaptured Sep 25, 2026
AI Evaluate scores each response against an AI-generated target answer and gives a detailed justification, and customers can supply their own evaluation criteria.
https://docs.xobin.com/en-us/article/what-is-xobin-ai-evaluate-1d3c0me/
- Public statementCaptured Sep 25, 2026
The help center presents the AI interviewer as a gate: recruiters only speak with the top performers who get past it.
https://docs.xobin.com/en-us/article/automating-interviews-what-is-an-ai-interview-in-xobin-16awmr1/
- Public statementCaptured Sep 25, 2026
The proctoring page says persistent tab-switch violations can trigger automatic disqualification, although reviewers can override verdicts in the session report.
Post-Market Monitoring
4 items38
- DocumentationCaptured Sep 25, 2026
The AI policy describes quarterly drift monitoring by candidate cohort, an AI incident log, incident SLAs (critical 24h, high 72h, medium 7d) and customer notification for material impacts. It cites a Q2 2024 drift in non-native English responses that was fixed through preprocessing changes.
- DocumentationCaptured Sep 25, 2026
Xobin says that if the model's score distribution shifts or its agreement with human raters slips, a re-validation gate stops continued production use and affected customers are notified.
- DocumentationCaptured Sep 25, 2026
The trust whitepaper says enterprise clients get quarterly SSDLC and compliance-posture reports and notices of critical CVEs, and cites a 17-hour fix of a critical OpenSSL CVE in 2024.
- AbsenceCaptured Sep 25, 2026
No working public status page or security contact was found. status.xobin.com returns a Cloudflare 525 error, Statuspage, Instatus, Better Stack and Freshstatus have no Xobin page, security.txt returns 404, and the xobin.com product-updates archive lists no posts.
Customer Documentation
6 items60
- DocumentationCaptured Sep 25, 2026
An ungated AI governance policy gives customers the candidate consent-checkbox wording, human-in-the-loop rules, fairness-testing approach, data residency options and a GDPR, EU AI Act, CCPA and DPDP compliance mapping.
- DocumentationCaptured Sep 25, 2026
A public help-center article explains AI interview score reports by must-have and nice-to-have skills and shows recruiters how to adjust a score manually with 'Click to Edit'.
https://docs.xobin.com/en-us/article/how-to-interpret-ai-interview-score-1ooa3mg/
- DocumentationCaptured Sep 25, 2026
The GDPR page says a Data Processing Addendum based on Model Contractual Clauses is available on request, that a DPO has been appointed and that DPIAs have been carried out.
- DocumentationCaptured Sep 25, 2026
The research hub publishes methodology pages on fairness and adverse impact (four-fifths rule, differential item functioning), assessment validation, proctoring and AI-bias auditing.
- Public statementCaptured Sep 25, 2026
The candidate FAQ tells candidates that Xobin does not really collect any data and that webcam access is required to start a test, and it says nothing about AI scoring or non-AI alternatives.
- AbsenceCaptured Sep 25, 2026
No NYC Local Law 144 or EU AI Act deployer guidance, public DPA text or subprocessor list was found on xobin.com, the GDPR page, the AI policy, the trust whitepaper or the help center.
Editorial notes
Company overview
Xobin is a Chennai-based talent-assessment company. Its Indian entity, Xobin Technologies Private Limited, was incorporated in March 2016, and it contracts in the US through Xobin, Inc. of Wilmington, Delaware. It claims 5,000+ customers in 55+ countries. On top of its test library it sells AI Evaluate, which uses generative AI to score written, audio and video answers against an AI-generated target answer and adds a written justification. It also sells agentic, guided and controlled AI interviews, marketed at about $5 per interview, and AI proctoring that uses face verification, gaze tracking and audio and device detection. Its newest product is NURA, an AI recruiting agent launched in July 2026 that parses resumes, runs AI interviews, scores candidates and hands a shortlist to recruiters. Xobin says its AI features run OpenAI models on Google Cloud in inference-only mode and are never trained on customer data.
Regulatory exposure
Xobin's own AI policy classifies its interview and scoring AI as high-risk under EU AI Act Annex III. The same policy still says the Act takes effect in 2026, but the Digital AI Omnibus has moved the Annex III obligations to 2 December 2027. The Article 5 bans already apply, including the ban on emotion recognition at work. Its communication assessments score the 'tone' of spoken answers, so Xobin should confirm that this scoring does not infer emotions. The AI interviewer and NURA filter candidates before a recruiter sees them, so NYC employers using them would likely need an independent Local Law 144 bias audit and candidate notice. Xobin publishes no such audit, and its fairness table has no race or ethnicity breakdown. In Illinois, employers must meet the HB 3773 notice duties, and the AI Video Interview Act likely covers AI-scored recorded interviews. Face-based identity checks in proctoring also raise biometric-privacy (BIPA) questions. California's FEHA rules on automated decision systems require four years of record retention, longer than the 12-month default Xobin's AI policy sets for assessments. Colorado's SB 26-189 disclosure regime starts on 1 January 2027.
Path to a higher score
Commission and publish an independent bias audit of AI interview scoring and NURA shortlisting that covers sex, race/ethnicity and intersectional groups, to replace the internal fairness table labelled 'illustrative'. Name the ISO 42001 certification body, certificate number and scope. Publish a system card for the AI scoring, a summary of the EU AI Act technical dossier, and deployer guidance that includes an Article 27 FRIA template and NYC LL 144 steps. Reconcile the three different retention periods it states (12 months, 7 years and 8 years) and publish a subprocessor list. Restore a working public status page, add a security.txt or vulnerability-disclosure contact and a model-change log. Finally, make NURA and proctoring defaults match the policy's mandatory human review, with no automatic rejection or disqualification until a recruiter confirms it.
Regulatory frame
What applies to assessment AI.
Assessment is the most-regulated surface in AI hiring. NYC LL 144, EEOC and OFCCP frameworks, Illinois HB 3773, and EU AI Act Annex III §4 all apply, with overlap on bias-audit obligations.
Compare
Xobin against its nearest-scoring peers.
The closest scores in Assessment AI, side by side.
- Side-by-sideXobin vs BrightHire51F
- Side-by-sideXobin vs SHL51F
- Side-by-sideXobin vs Test Partnership50F
- Side-by-sideXobin vs Versant by Pearson49F
Ranking
Others rated in Assessment AI
Ranked 21 of 82 by weighted total under rubric v1.0. The ordering is arithmetic on the rubric and carries no view on which tool suits a given hiring process.
- #18Glider AI52F
- #19BrightHire51F
- #20SHL51F
- #21XobinThis profile50F
- #22Test Partnership50F
- #23Versant by Pearson49F
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. See the full policy on About.
Casework has no commercial relationship with this vendor.