Coderbyte
HireAIScore rates Coderbyte 42 out of 100 (grade F, Substantial gaps) under rubric v1.0, last reviewed September 25, 2026. Coderbyte ranks 56 of 82 vendors rated in Assessment AI, against a category average of 46. That score is drawn from 31 evidence items across 7 rubric criteria for Coderbyte — 26 with a cited source, 5 recording that nothing was located.
Score breakdown
Coderbyte, 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
44
+8.8
Bias Audit Transparency
Weight 18% · category avg 41
20
+3.6
FRIA Support
Weight 15% · category avg 31
33
+5.0
Data Governance Disclosure
Weight 15% · category avg 53
56
+8.4
Human Oversight Design
Weight 12% · category avg 55
60
+7.2
Post-Market Monitoring
Weight 12% · category avg 38
40
+4.8
Customer Documentation
Weight 8% · category avg 58
58
+4.6
Total weighted score
42.4(rounded to 42 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
Human Oversight Design
60raw / 100
Contributes 7.2 to the total. 60 against a 55 category average.
Weakest criterion
Bias Audit Transparency
20raw / 100
Contributes 3.6 to the total. 20 against a 41 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
- 31
- Documentation
- 21
- Audit report
- 2
- Public statement
- 3
- Absence
- 5
- With a source URL
- 26 of 31
- Source hostnames
- 7
- Fewest items
- 3
- Bias Audit Transparency
These figures measure how thoroughly Coderbyte was reviewed, not how Coderbyte performed — an absence row, recording that nothing was located, is counted like any other item.
Article 11 Technical Documentation
4 items44
- DocumentationCaptured Sep 25, 2026
The AI features page lists six opt-in AI features and gives a data-flow diagram for each, showing resume and phone-screen evaluation on AWS Bedrock, phone-screen audio handled by ElevenLabs, and 'no retention / no training' labels.
- DocumentationCaptured Sep 25, 2026
A 'Test validity and science' page cites EEOC guidance and describes a five-step content-validation process for library challenges, overseen by a Head of Assessment Science, but it covers neither the AI graders nor customer-built challenges.
- DocumentationCaptured Sep 25, 2026
The AI Terms Addendum (last updated 16 May 2026) warns that output may be inaccurate and that AI 'can perpetuate biases', and it names ElevenLabs as the phone-screen sub-processor.
- AbsenceCaptured Sep 25, 2026
No model cards, no explanation of how resume match scores, phone-screen final scores or 'confidence scores' are computed, no ISO 42001 claim and no instructions for use turned up in the 140-article help center sitemap, the trust center, the AI Terms or the homepage FAQ.
Bias Audit Transparency
3 items20
- DocumentationCaptured Sep 25, 2026
Coderbyte says it collects demographic data from poorly performing beta candidates to check library challenges for adverse impact before full release, but it publishes no results, method, sample sizes or auditor, and the step does not cover AI grading, resume grading or phone screens.
- Public statementCaptured Sep 25, 2026
The homepage FAQ says library questions are 'rigorously tested for fit and bias' but that Coderbyte offers no off-the-shelf assessments that can be validated, and it offers validity-study support only on Enterprise.
- AbsenceCaptured Sep 25, 2026
No NYC Local Law 144 bias audit or independent fairness audit of any Coderbyte AI feature turned up on coderbyte.com, the help center, the Drata trust center (full document list checked in a browser) or Warden AI (trust.warden-ai.com/coderbyte returns 404).
FRIA Support
4 items33
- Public statementCaptured Sep 25, 2026
Coderbyte advises organizations with EU AI Act or high-risk concerns not to enable AI features 'unless and until' both parties agree the requirements have been addressed, which is a caution rather than deployer guidance.
https://help.coderbyte.com/knowledge/coderbyte-security-checklist
- DocumentationCaptured Sep 25, 2026
A public GDPR DPIA, titled 'Coderbyte DPIA 2025 (Incomplete)' in its file metadata and last reviewed on 16 June 2025, calls itself a 'Good Practice' DPIA because the processing is 'not likely to result in a high risk'.
https://insights.coderbyte.com/hubfs/Security/Coderbyte%20DPIA.pdf
- DocumentationCaptured Sep 25, 2026
The Drata trust center lists a Data Privacy Impact Assessment, an application penetration test and an architecture diagram as request-access documents.
- AbsenceCaptured Sep 25, 2026
No Article 27 FRIA template, Article 26 deployer-obligation guidance or instructions for use turned up in the help center, AI Terms, DPA or trust center.
Data Governance Disclosure
5 items56
- Audit reportCaptured Sep 25, 2026
Sensiba LLP's SOC 2 Type 2 report, signed 25 June 2026 and publicly downloadable, covers the Security criterion for 16 May 2025 to 15 May 2026, with a clean opinion and no exceptions noted.
- DocumentationCaptured Sep 25, 2026
Coderbyte states that candidate and assessment data are deleted within 90 days of a subscription going inactive, that screen and video recordings are kept for up to 365 days, and that customers can use assessments without collecting PII.
https://help.coderbyte.com/knowledge/coderbyte-security-checklist
- DocumentationCaptured Sep 25, 2026
The public sub-processor list (updated 21 September 2026) names AWS, including 'Bedrock for AI features', plus Veriff for identity verification and Session Rewind for activity recording.
- DocumentationCaptured Sep 25, 2026
In the AI Terms, Coderbyte says its contracts bar covered AI sub-processors from training on or retaining customer data, and phone-screen audio sent to ElevenLabs counts as de-identified data outside the DPA's sub-processor notice procedure; ElevenLabs is not on the public sub-processor list.
- DocumentationCaptured Sep 25, 2026
By default, candidate reports can be shared with anyone, whether or not they have a Coderbyte account, until an admin restricts visibility.
https://help.coderbyte.com/knowledge/enabling-private-reports
Human Oversight Design
5 items60
- DocumentationCaptured Sep 25, 2026
Every AI feature stays off until the organization accepts the AI Terms and opts in, and Coderbyte says some AI automations are withheld 'to ensure a human is always involved in hiring decisions'.
- DocumentationCaptured Sep 25, 2026
Phone-screen reports include the audio recording and full transcript, and for each criterion they show whether it was met, a confidence score and supporting evidence from the candidate's answers.
https://help.coderbyte.com/knowledge/phone-screen-report-overview
- DocumentationCaptured Sep 25, 2026
ATS automations triggered by a phone screen or resume grade take no score conditions and there is no reject action, but completed assessments can advance candidates automatically on qualification score and cheating-detection result.
- DocumentationCaptured Sep 25, 2026
The changelog records features that let recruiters override AI-generated resume criteria (May 2026) and override challenge scores in any candidate report (August 2026).
- DocumentationCaptured Sep 25, 2026
The AI report-analysis article says candidate submissions 'should always be reviewed by domain experts before any decision is made'.
https://help.coderbyte.com/knowledge/ai-powered-candidate-report-analysis
Post-Market Monitoring
5 items40
- DocumentationCaptured Sep 25, 2026
An Uptime.com status page linked from the site footer tracks a single 'Website' component, and in a browser check its history view listed no past incidents or maintenance.
- DocumentationCaptured Sep 25, 2026
A public monthly changelog (entries from March to August 2026 visible) records AI changes such as new grading features, the phone-screen launch and model updates to the candidate-facing AI assistant, but not version changes or evaluation results for the grading models.
- Audit reportCaptured Sep 25, 2026
The SOC 2 report confirms a process for external users to report security failures, incidents and complaints, and says no significant incidents affected customers in the review period.
- DocumentationCaptured Sep 25, 2026
The trust center's policy list includes an Incident Response Plan and a Responsible Disclosure Policy.
- AbsenceCaptured Sep 25, 2026
No monitoring of AI accuracy, drift or fairness, no AI-incident reporting channel and no security.txt (coderbyte.com/.well-known/security.txt returns 403) turned up.
Customer Documentation
5 items58
- DocumentationCaptured Sep 25, 2026
The public DPA (linked from coderbyte.com/dpa) includes EU and UK standard contractual clauses and lists 'automated decision making and profiling' and 'webcam recordings' within the processing it covers.
- DocumentationCaptured Sep 25, 2026
Help-center articles walk deployers through each AI feature, for example explaining that AI project grading scores custom rubric criteria and normalizes them to a 0-10 project score.
- Public statementCaptured Sep 25, 2026
The homepage FAQ links the Terms, Privacy Policy, DPA and DPIA, and offers Enterprise customers hands-on information-security reviews, validity-study support and customizable data residency.
- DocumentationCaptured Sep 25, 2026
The personality-testing article tells customers that the tests 'are not peer-reviewed science and are not recommended for use during hiring processes'.
- AbsenceCaptured Sep 25, 2026
No guidance on NYC Local Law 144, Illinois, Colorado or EU AI Act deployer duties and no candidate-notice template for AI phone screens or resume grading turned up, and the AI Terms assign all notices and consents to the customer.
Editorial notes
Company overview
Coderbyte Enterprise Inc. was founded in June 2012 and gives its address as Carlstadt, New Jersey. Its SOC 2 description puts its customer base at 3,000+ organizations hiring for technical roles. The core product is coding assessments graded automatically against test cases, plus live coding interviews. The AI features are optional. Each one is switched on per organization after it accepts an AI Terms Addendum. They grade take-home projects and free-form answers against editable rubrics, score resumes against AI-generated job criteria, run agentic phone screens (AWS Bedrock does the evaluation and ElevenLabs handles the audio) and add webcam proctoring with eye tracking, face detection and deepfake detection. Pricing starts at $99 a month or $1,799 a year. Enterprise plans start at $12,495 a year.
Regulatory exposure
Resume grading, agentic phone screens and AI grading of candidate work analyse applications and evaluate candidates, which puts them in the EU AI Act's Annex III point 4 high-risk category. The Digital AI Omnibus moved those duties to 2 December 2027 but did not change them. Coderbyte advises customers with EU AI Act concerns not to switch the AI features on, while its public DPIA calls its processing 'not likely to result in a high risk'. On NYC Local Law 144, the AI Terms bar customers from using AI output as a 'substantial factor' in consequential decisions. But the products output match scores and final scores built to prioritize candidates, and Coderbyte has published no bias audit. DCWP has promised tougher enforcement since the December 2025 Comptroller audit. Illinois has required notice of AI use in employment decisions since 1 January 2026, and video responses plus AI webcam proctoring may also bring in the AI Video Interview Act. The AI Terms leave every notice and consent to the customer. Colorado's SB 26-189 disclosure regime starts on 1 January 2027. California's FEHA rules require employers to keep records for four years, longer than Coderbyte keeps them: it deletes candidate data 90 days after a subscription lapses.
Path to a higher score
Coderbyte already publishes a downloadable SOC 2 Type 2 report, a DPIA and a data-flow diagram for each AI feature, so most of the gap is AI-specific evidence. The weakest category would rise most with an independent bias audit of resume grading, phone screens and AI project grading, plus a public summary in the Local Law 144 format and the adverse-impact results its validity page says it already collects. Technical documentation and data governance need model documentation for each AI feature: which Bedrock models it uses, what the confidence scores mean, its known limits, and whether names, contact details and zip codes are stripped from resumes. The remaining categories need an updated DPIA that treats resume and phone-screen scoring as high-risk, EU deployer guidance with a FRIA template in place of the advice not to use the features, candidate-notice templates for AI phone screens, and a public log of AI incidents and model changes.
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
Coderbyte against its nearest-scoring peers.
The closest scores in Assessment AI, side by side.
- Side-by-sideCoderbyte vs Talview42F
- Side-by-sideCoderbyte vs Talogy42F
- Side-by-sideCoderbyte vs Tengai42F
- Side-by-sideCoderbyte vs Mercer | Mettl42F
Ranking
Others rated in Assessment AI
Ranked 56 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.
- #53Wonderlic43F
- #54CodeSignal43F
- #55Talview42F
- #56CoderbyteThis profile42F
- #57Talogy42F
- #58Tengai42F
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.