Candidate Screening Agent - Title VII, EU AI Act Annex III, NYC AEDT | Gosign
From US EEOC Four-Fifths Rule plus OFCCP Internet Applicant Rule plus NYC Local Law 144 AEDT bias audit through UK Equality Act 2010 to EU AI Act 2024 Annex III(4)(a) High-Risk AI System with Article 14 human oversight plus Article 13 transparency plus conformity assessment - one auditable candidate screening pipeline across CV-parsing plus resume-screening plus shortlist-generation plus bias-monitoring plus EU Pay Transparency Directive 2023/970 compliance.
AI-bias-audited CV screening: Title VII/EEOC Four-Fifths Rule, EU AI Act 2024 Annex III(4)(a) high-risk AI conformity, NYC Local Law 144 AEDT and UK Equality Act 2010 - Article 14 human oversight.
Analyse your processAuswahl aus über 5.000 Projekten in 25 Jahren Softwareentwicklung
US Title VII plus ADEA plus EU AI Act 2024 Annex III(4)(a) plus UK Equality Act 2010 plus EU Pay Transparency Directive 2023/970 - one auditable candidate-screening pipeline across CV-parsing plus resume-screening plus shortlist-generation plus AI-bias-audit
The Agent decomposes the screening process into 13 documented decision steps with a defined decider per step (rules engine, AI agent, human) and per-criterion explainability replacing black-box scoring. Knockout criteria run deterministically through a rules engine (US: work authorisation under I-9 plus E-Verify; UK: Right to Work plus Skilled Worker visa; EU: Single Permit Directive). Profile matching runs through AI extraction with individually justified partial scores per requirement, never one composite score. Bias monitoring runs continuously against EEOC Four-Fifths Rule plus statistical-significance testing per EEOC Uniform Guidelines 29 CFR Part 1607. The output is an auditable shortlist that satisfies EU AI Act Articles 12, 13, 14 by construction, NYC Local Law 144 AEDT annual bias audit, OFCCP Internet Applicant Rule 41 CFR 60-1.12 recordkeeping, plus UK Equality Act 2010 objective-justification standard.
Outcome: For 200 to 800 applications per role across UK plus EU plus US, the Agent produces an auditable shortlist instead of a black-box score, complete decision documentation per candidate satisfying EU AI Act Articles 12 to 14 from the 2 August 2026 deadline, NYC Local Law 144 annual independent third-party bias audit, OFCCP federal contractor Internet Applicant recordkeeping, EEOC EEO-1 plus VETS-4212 reporting feed, plus GDPR Article 22 challengeability with per-criterion rationale enabling candidate dispute. Time-to-shortlist compressed from 3-6 weeks to under 1 week for 5,000-employee multinational; auditor finding rate from typical 8-15% on AI screening to under 2% with conformity-assessed pipeline; (UK: structured-interview consistency provides direct tribunal-defence; US: Four-Fifths Rule auto-pause prevents disparate-impact accumulation).
The 13 deterministic candidate-screening steps span US Title VII plus EEOC plus OFCCP plus NYC Local Law 144 plus UK Equality Act 2010 plus EU AI Act 2024 plus GDPR plus EU Pay Transparency Directive 2023/970 - and precisely because each step is determined by statute, regulation, or standard, the pipeline is machine-reproducible plus audit-defensible:
August 2026: EU AI Act Annex III(4)(a) deadline plus EEOC Four-Fifths Rule plus NYC Local Law 144 AEDT audit plus UK Equality Act 2010 plus EU Pay Transparency Directive 2023/970 - one auditable candidate screening pipeline
International candidate screening does not run on one regulatory standard - it runs on six overlapping regimes simultaneously across UK + EU + US. From 2 August 2026, every AI system that filters job applications across the EU is a High-Risk AI System under EU AI Act Article 6 plus Annex III(4)(a). Organisations that cannot demonstrate Article 16 conformity assessment plus Article 47 EU declaration of conformity plus Article 48 CE marking plus Article 26(7) worker-representative information plus Article 27 fundamental rights impact assessment plus Article 14 human oversight plus Article 12 record-keeping by then will have to shut down the automation. Not at some point. On a fixed date.
A US-headquartered multinational with 5,000 employees running 200-800 applications per role concurrently faces US Title VII Civil Rights Act 1964 plus ADEA plus ADA plus Equal Pay Act plus GINA plus EEOC Four-Fifths Rule under Uniform Guidelines on Employee Selection Procedures 29 CFR Part 1607, US OFCCP Internet Applicant Rule 41 CFR 60-1.12 plus Executive Order 11246 plus Section 503 plus VEVRAA, US NYC Local Law 144 of 2021 AEDT bias audit plus Illinois HB 645 plus Maryland HB 1202 plus California SB 1162, UK Equality Act 2010 plus EHRC Code of Practice on Employment plus Section 60 health-enquiry prohibition, EU GDPR Article 22 automated decision-making per CJEU C-634/21 SCHUFA judgment plus Article 35 DPIA plus Article 88, EU AI Act 2024 Article 6 plus Annex III(4)(a), plus EU Pay Transparency Directive 2023/970 with 7 June 2026 transposition deadline.
US Title VII plus ADEA plus EU AI Act 2024 Annex III(4)(a) plus UK Equality Act 2010 plus EU Pay Transparency Directive 2023/970 - one auditable candidate-screening pipeline
This Agent follows the Decision Layer principle: each decision is either rule-based, AI-assisted, or explicitly assigned to a human - with per-criterion explainability replacing black-box scoring.
The obvious challenge is familiar: 200, 400, sometimes 800 applications per role. Recruiters who, after the fiftieth CV, no longer apply the same standards as for the first. Hiring managers who ask after three weeks why the shortlist is not ready.
The real problem runs deeper. Most organisations using AI in screening do not know how their algorithms evaluate. They do not know the weighting. They cannot explain why Candidate A is on the shortlist and Candidate B is not. That is precisely where risk accumulates - and where each jurisdiction now demands documented architecture.
The Mobley v. Workday case makes this tangible. An applicant sues not the prospective employer but the software vendor whose AI screened him out. The US federal court certifies the claim as a class action - alleging systematic discrimination by age, ethnicity, and disability. A University of Washington study shows: in AI-driven CV screenings, names associated with white ethnicity were preferred in 85% of cases. In some occupational groups, Black male applicants were disadvantaged in 100% of test cases. (US: this exposes vendors plus deployers under Title VII Section 703(k) burden-shifting; UK: equivalent exposure under Equality Act 2010 indirect-discrimination test).
Every screening failure carries direct cost. In the US, EEOC charges trigger conciliation plus litigation plus consent decrees with monetary relief plus injunctive relief plus monitoring; OFCCP Predetermination Notices plus Notices of Violation trigger Conciliation Agreements plus debarment from federal contracts; NYC Local Law 144 violations trigger USD 500-1,500 daily-accruing penalties; Illinois HB 645 plus Maryland HB 1202 plus California SB 1162 plus Colorado AI Act create overlapping state-level liability. In the UK, Equality Act tribunal awards include uncapped compensation plus aggravated plus exemplary damages plus interest plus reinstatement plus declaration; ICO penalties up to GBP 17.5 million or 4% of global turnover. In the EU, AI Act penalties up to EUR 35 million or 7% of worldwide annual turnover for prohibited practices, EUR 15 million or 3% for high-risk AI provider obligation breach, EUR 7.5 million or 1% for misleading information; plus GDPR penalties up to EUR 20 million or 4% global turnover plus CJEU SCHUFA Article 22 enforcement.
13 deterministic candidate-screening steps span US Title VII plus EEOC plus OFCCP plus NYC Local Law 144 plus UK Equality Act 2010 plus EU AI Act 2024 plus EU Pay Transparency Directive
Unlike single-jurisdiction screening (8-10 steps), cross-jurisdictional candidate screening requires 13 deterministic steps because of regulatory overlap: EU AI Act conformity-assessment gate plus GDPR lawful-basis validation plus document parsing with protected-characteristic redaction plus rule-based knockout (work authorisation, licensure, security clearance) plus per-criterion AI matching with explainability plus continuous bias monitoring per Four-Fifths Rule plus adverse-impact escalation plus per-candidate decision record plus recruiter shortlist review with structured-interview kit plus decision documentation with neutrality attestation plus downstream synchronisation plus EU Pay Transparency plus periodic AEDT bias audit reporting.
Concrete cross-border scenario: US-HQ S&P 500 manufacturer, 5,000 employees (3,200 US in 14 states including 250 NYC roles, 1,200 UK, 600 EU), 280 active requisitions, 65,000 applications annually. Outputs: 65,000 EU AI Act decision records, 250 NYC Local Law 144 annual AEDT bias audit submissions, 3,200 EEO-1 Component 1 entries, 280 OFCCP Internet Applicant logs, 1,200 UK Equality Act tribunal-defence dossiers, 600 GDPR Article 22 challenge-mechanism records, plus quarterly bias-monitoring reports plus annual conformity-assessment refresh.
In the Decision Layer, 7 of 13 steps are rule-engine decisions (tier R) - EU AI Act conformity gate, GDPR lawful-basis validation, knockout criteria, adverse-impact escalation, decision record generation, EU Pay Transparency processing, periodic AEDT audit reporting. 4 of 13 steps are AI-augmented (tier A) - document parsing with redaction, per-criterion matching, bias monitoring, downstream synchronisation. 2 of 13 steps are human-judgement (tier H) - recruiter shortlist review with structured-interview kit, decision documentation with rationale plus protected-characteristic neutrality attestation. Every step is documented with timestamp, decider type, rationale, plus challenge mechanism per GDPR Article 22 plus EU AI Act Article 13.
CV-parsing, knockout-criteria, AI-matching, bias-audit, EU AI Act conformity differentiate screening from audit/compliance
The 6 candidate-screening dimensions distinguish this Agent from generalised HR audit support: (1) CV-parsing plus document extraction with provenance tracking plus protected-characteristic redaction (name, photo, age, gendered pronouns, alma mater proxies); (2) rule-based knockout criteria for work authorisation (US: I-9 plus E-Verify; UK: Right to Work plus Skilled Worker visa; EU: Single Permit Directive) plus regulated-profession licensure plus security clearance; (3) per-criterion AI matching with individually justified partial scores plus weighting rationale (never one composite black-box score); (4) continuous bias monitoring per EEOC Four-Fifths Rule plus statistical-significance testing plus Hazelwood standard-deviation analysis; (5) EU AI Act 2024 Annex III(4)(a) High-Risk conformity assessment plus CE marking plus FRIA plus Article 14 human oversight plus Article 12 record-keeping; (6) NYC Local Law 144 annual independent third-party bias audit plus 10-business-day candidate notice plus public posting.
The architecture satisfies EU AI Act Articles 12, 13, and 14 by construction, not retrofit. Article 14 requires human oversight - no committee can manually review 400 applications and simultaneously exercise oversight, but a committee can review a shortlist with documented per-criterion assessments and can trace why Candidate A scored 89 on competence fit and Candidate B scored 61. Article 13 requires transparency - a rejected applicant can understand at which named requirement the application failed, not at an aggregate score that explains nothing. Article 12 requires records that enable subsequent evaluation - the Decision Log captures every step with timestamp, decider type, rationale, satisfying this not as by-product but as core function.
Cross-system integration with Workday + SAP + Oracle + Greenhouse + Lever + iCIMS + Eightfold + HireVue + Personio
The Agent integrates with the full global recruiting stack: Workday Recruiting plus Talent Optimization plus Skills Cloud, SAP SuccessFactors Recruiting plus Recruiting Marketing plus Onboarding, Oracle Recruiting Cloud plus Talent Acquisition Cloud, ADP Recruiting Management. For dedicated ATS: Greenhouse Recruiting plus Greenhouse Inclusion plus Greenhouse Connect, Lever ATS plus LeverTRM, iCIMS Talent Cloud, SmartRecruiters, Workable, Recruitee, JazzHR, BreezyHR, Bullhorn ATS, Avature, Lattice Recruiting, Personio Recruiting (DACH plus EU mid-market), BambooHR ATS. For AI talent intelligence: Eightfold AI Talent Intelligence, Beamery, HiringSolved, Hiretual, Pymetrics. For video interviewing plus assessment: HireVue Video Interviewing, Modern Hire (HireVue), Plum. For executive search plus RPO: Korn Ferry, Heidrick & Struggles, Russell Reynolds, Mercer, Aon, Willis Towers Watson. The Candidate Screening Agent is the litmus test for high-risk governance readiness - if an organisation can deploy this Agent with full EU AI Act compliance, it can deploy any Annex III high-risk agent.
Micro-Decision Table
Who decides in this agent?
13 decision steps, split by decider
Assess EU AI Act 2024 Article 6 plus Annex III(4)(a) High-Risk classification plus conformity assessment status before processing application Is the screening system classified as High-Risk AI System under EU AI Act Annex III(4)(a) for recruitment/selection, has the provider completed Article 16 conformity assessment under Annex VI internal control or Annex VII third-party route, has the EU declaration of conformity been issued under Article 47, has the CE marking been affixed under Article 48, has the deployer completed Article 27 fundamental rights impact assessment, has the deployer informed worker representatives under Article 26(7)? Rules Engine Auditor
Deterministic compliance gate per EU AI Act 2024 Article 6 plus Annex III(4)(a) plus Article 16 conformity assessment plus Article 47 declaration plus Article 48 CE marking plus Article 26(7) deployer obligations plus Article 27 fundamental rights impact assessment with 2 August 2026 deadline; without conformity assessment plus CE marking plus FRIA plus worker representative consultation the system cannot lawfully be deployed for recruitment/selection in the EU market
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by: Auditor
Validate lawful basis under GDPR Article 6 plus Article 9 plus Article 22 plus consent or contract-necessity for automated processing What is the lawful basis for processing each candidate's personal data: Article 6(1)(b) pre-contractual measures for shortlist generation, Article 6(1)(f) legitimate interests for talent-pipeline development with documented LIA balancing test, Article 9(2)(b) employment-context special-category-data exception under Member State law for diversity monitoring (where collected), plus Article 22 explicit consent or contract necessity for any solely-automated decision with suitable safeguards including human intervention plus right to express view plus right to contest; plus EU Pay Transparency Directive 2023/970 prohibition on pay-history questions and (UK: Section 60 Equality Act 2010 prohibition on pre-employment health enquiries except intrinsic-function inquiries) Rules Engine Auditor
Deterministic lawful-basis classification per GDPR Articles 6, 9, 22 plus EDPB Guidelines on Article 22 plus CJEU C-634/21 SCHUFA judgment treating credit-score-style profiling as Article 22 decision; plus EU Pay Transparency Directive 2023/970 transposition deadline 7 June 2026; plus UK ICO Guidance on AI and Data Protection 2023; plus Member State derogations under Article 88 including German BDSG Section 26
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by: Auditor
Parse application documents and extract structured candidate data with provenance tracking Extract structured data from CV, cover letter, application form, LinkedIn profile, portfolio attachments using deterministic document-parsing plus AI extraction with provenance tracking per field (source document plus location plus extraction confidence) plus mandatory exclusion of GINA-prohibited genetic information plus protected-characteristic markers (age proxies, race proxies, disability indicators, religious affiliation, sexual orientation, marital status, parental status) per Title VII plus ADA plus ADEA plus GINA plus UK Equality Act 2010 nine protected characteristics AI Agent
AI-driven document parsing with deterministic protected-characteristic redaction; AI handles unstructured-document extraction across CV formats plus cover letter prose plus LinkedIn data; deterministic redaction layer removes name, photo, age, address proxies, alma mater proxies, gendered pronouns before downstream matching to mitigate disparate-impact risk under EEOC Four-Fifths Rule
Decision Record
Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.
Challengeable by:
Apply rule-based knockout criteria for formal completeness and statutory minimum qualifications What knockout criteria apply: required work authorisation status (US I-9 plus E-Verify or UK Right to Work plus Skilled Worker visa or EU Member State equivalent), regulated-profession licensure status (medical, legal, accounting, engineering), statutory minimum age, security clearance level (DoD Secret/Top Secret, UK SC/DV, EU Cosmic Top Secret), language requirements documented as bona fide occupational requirement under Title VII plus UK Equality Act objective justification? Rules Engine
Deterministic knockout per rule-engine, no AI judgement; bona fide occupational qualifications under Title VII Sec 703(e) plus UK Equality Act objective justification test plus EU equivalent must be documented plus minimally restrictive; work-authorisation per Immigration and Nationality Act plus UK Immigration Act 2016 plus EU Single Permit Directive plus Long-Term Residence Directive; licensure per state-board plus UK regulator plus EU Member State professional-qualifications recognition under Directive 2005/36/EC
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by:
Match qualifications and experience to requirement profile with per-criterion weighted scoring and individual rationale For each requirement-profile criterion (technical skills, professional experience years, industry experience, education level, certifications, language proficiency, project portfolio), what is the candidate's match score (0-100) with documented rationale citing specific CV evidence plus weighting rationale? AI provides per-criterion partial scores with explainable rationale, never a single black-box composite; weights validated against bona fide occupational qualifications and reviewed quarterly for disparate-impact risk AI Agent
AI-driven semantic matching with per-criterion explainability addresses EU AI Act Article 13 transparency plus Article 14 human oversight plus EEOC Uniform Guidelines validation requirements plus UK Equality Act objective-justification test for indirect discrimination; per-criterion rationale enables candidate challenge under GDPR Article 22 right to obtain human intervention plus right to express view plus right to contest decision plus right to explanation
Decision Record
Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.
Challengeable by:
Run continuous bias monitoring with EEOC Four-Fifths Rule plus Adverse Impact Ratio plus statistical significance testing Across the candidate pool, what is the selection rate per protected group (race/ethnicity per EEO-1 categories, sex, age 40+, disability, veteran status), what is the Impact Ratio (selection rate of group / selection rate of highest-rate group), does any protected group fall below 0.80 (Four-Fifths Rule threshold under EEOC Uniform Guidelines), and is the disparity statistically significant (Fisher's Exact Test, Chi-Square, Z-test for two proportions, with p<0.05 standard)? AI Agent Auditor
Statistical fairness analysis per EEOC Uniform Guidelines on Employee Selection Procedures 29 CFR Part 1607 plus Four-Fifths Rule plus standard deviation tests Hazelwood School District v United States 1977; plus NYC Local Law 144 AEDT annual independent third-party audit covering selection rate plus impact ratio; plus EU AI Act Article 10 data quality plus Article 15 accuracy and bias detection; thresholds configured per jurisdiction
Decision Record
Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.
Challengeable by: Auditor
Escalate adverse-impact findings to compliance team with Notice of Violation prevention workflow If Four-Fifths Rule or statistical-significance threshold is breached, what escalation: pause shortlist generation, alert compliance team plus DPO plus EU AI Act human-oversight controller, generate adverse-impact report with affected protected group plus impact ratio plus statistical significance plus business-necessity review plus less-discriminatory-alternative analysis under Griggs v Duke Power Co plus Title VII Section 703(k) burden-shifting framework? Rules Engine Auditor
Threshold-based alerting per Title VII disparate-impact framework plus EEOC Uniform Guidelines plus NYC Local Law 144 audit-finding remediation plus EU AI Act Article 17 quality management system plus Article 79 corrective action; auto-pause prevents continued use of biased system pending business-necessity review plus less-discriminatory-alternative analysis
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by: Auditor
Generate per-candidate decision record with pyramidopener context and challenge mechanism For each candidate, generate decision record with: candidate ID plus job ID plus timestamp, applied filters plus knockout-criteria results, per-criterion match scores plus rationale, AI confidence per criterion, bias-monitoring snapshot, recruiter override flag (if any), final decision (advance to interview, reject, hold), GDPR Article 13/14 transparency notice plus Article 22 challenge mechanism plus 10-business-day NYC Local Law 144 candidate notice (NYC roles), plus retention schedule per OFCCP Internet Applicant Rule (2 years federal contractor) plus UK Equality Act 6 months claim window plus EU GDPR storage limitation Rules Engine
Deterministic decision-record generation under EU AI Act Article 12 record-keeping plus Article 13 transparency plus GDPR Article 13/14 information rights plus Article 22 right to contest plus OFCCP Internet Applicant Rule recordkeeping plus NYC Local Law 144 candidate-notice requirement plus EEOC EEO-1 plus VETS-4212 reporting feed; record format harmonised for cross-jurisdiction audit
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by:
Recruiter reviews shortlist with structured interview kit plus blinded comparator panels Recruiter evaluates each shortlisted candidate against requirement profile using structured-interview kit (consistent questions across all candidates per Title VII plus UK Equality Act objective-justification standard), blinded comparator review (rationale per criterion vs cohort distribution), plus override authority where AI scoring conflicts with recruiter assessment with documented override rationale; (UK: structured-interview consistency required as defence to indirect-discrimination claim under Equality Act 2010); (US: Uniform Guidelines validation requires consistent application of selection procedures) Human
Human decision required under EU AI Act Article 14 human oversight plus GDPR Article 22 prohibition on solely automated decisions plus Title VII plus UK Equality Act 2010 plus structured-interview methodology validated by Schmidt and Hunter 1998 meta-analysis showing structured interviews predict job performance more reliably than unstructured (r=0.51 vs r=0.38); the Agent provides structure for recruiter judgement, never the verdict
Decision Record
Challengeable: Yes - via manager, works council, or formal objection process.
Challengeable by:
Document recruiter decision with rationale and protected-characteristic neutrality attestation Recruiter documents the screening decision with: decision (advance, reject, hold), rationale citing specific job-requirement criteria plus interview evidence, override rationale if differing from AI score, attestation of protected-characteristic neutrality plus structured-interview consistency, plus consideration of less-discriminatory alternatives where adverse-impact alert was raised; documentation persisted in immutable Decision Log per EU AI Act Article 12 plus Article 26(6) deployer obligation to keep automatically-generated logs for at least 6 months Human
Mandatory documentation per EU AI Act Article 12 record-keeping plus Article 26(6) deployer logs plus EEOC Uniform Guidelines plus OFCCP Internet Applicant Rule plus UK Equality Act 2010 evidential burden in tribunal claims plus GDPR Article 22 challengeability plus Section 703(k) burden-shifting under Title VII
Decision Record
Challengeable: Yes - via manager, works council, or formal objection process.
Challengeable by:
Synchronise final decision with downstream ATS, EEO/OFCCP reporting, and candidate-relationship management Has the screening decision been transmitted to: ATS candidate status (Workday Recruiting, SAP SuccessFactors Recruiting, Oracle Recruiting Cloud, Greenhouse, Lever, iCIMS, Personio Recruiting), EEO-1 plus VETS-4212 plus OFCCP Internet Applicant recordkeeping feed, NYC Local Law 144 audit-data feed (NYC roles), candidate-relationship-management for talent-pool development with explicit consent under GDPR Article 6(1)(a), plus rejection notification with GDPR Article 13/14 transparency information plus Article 22 challenge mechanism plus 10-business-day NYC Local Law 144 notice (NYC roles)? AI Agent Auditor
Automated downstream synchronisation via SCIM plus REST API plus SFTP file feeds; AI surfaces synchronisation failures for human review without auto-correcting decision errors; integration tested for cross-jurisdiction recordkeeping consistency
Decision Record
Challengeable: Yes - fully documented, reviewable by humans, objection via formal process.
Challengeable by: Auditor
Process EU Pay Transparency Directive 2023/970 plus EU Mobility Directive plus cross-border applicant data For EU plus UK applicants, what cross-border compliance applies: EU Pay Transparency Directive 2023/970 prohibition on pay-history question plus mandatory pay-disclosure in vacancy notice or before interview, gender-neutral job titles plus job descriptions, plus burden-of-proof reversal in pay-discrimination claims; (UK: equivalent pay-transparency under Equality Act 2010 plus Equal Pay Act sex-comparator framework plus Gender Pay Gap reporting); plus cross-border data transfer assessment under GDPR Chapter V plus UK GDPR Schedule 21 plus US-UK Data Bridge plus EU-US Data Privacy Framework adequacy? Rules Engine
Deterministic cross-border compliance per EU Pay Transparency Directive 2023/970 transposition deadline 7 June 2026 plus UK Equality Act 2010 plus Gender Pay Gap reporting plus GDPR Chapter V transfer rules plus UK GDPR plus EU-US Data Privacy Framework Commission Decision 2023/1795 plus US-UK Data Bridge Extension
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by:
Generate periodic AEDT bias audit report and EEO-1 plus VETS-4212 plus OFCCP submissions What periodic compliance reporting is generated: NYC Local Law 144 annual independent third-party AEDT bias-audit report covering selection rate plus impact ratio across sex, race/ethnicity, intersectional EEOC Component 1 categories with public posting plus 10-business-day candidate notice; plus EEO-1 Component 1 annual filing (employers 100+ or federal contractors 50+) by deadline; plus VETS-4212 protected-veteran annual filing; plus OFCCP Internet Applicant Rule recordkeeping (2 years federal contractor); plus EU AI Act Article 12 record-keeping plus Article 26(6) deployer logs (6 months minimum)? Rules Engine Auditor
Deterministic periodic-compliance reporting per NYC Local Law 144 annual audit deadline plus EEO-1 deadline plus VETS-4212 deadline plus OFCCP recordkeeping retention plus EU AI Act Article 12 plus Article 26(6); third-party audit firm engagement required for NYC Local Law 144 independence under Department of Consumer and Worker Protection rule
Decision Record
Challengeable: Yes - rule application verifiable. Objection possible for incorrect data or wrong rule version.
Challengeable by: Auditor
Decision Record and Right to Challenge
Every decision this agent makes or prepares is documented in a complete decision record. Affected employees can review, understand, and challenge every individual decision.
Does this agent fit your process?
We analyse your specific HR process and show how this agent fits into your system landscape. 30 minutes, no preparation needed.
Analyse your processGovernance Notes
Assessment
Prerequisites
- Cloud HCM-embedded or dedicated ATS with API access: Workday Recruiting plus Talent Optimization plus Skills Cloud, SAP SuccessFactors Recruiting plus Recruiting Marketing plus Onboarding, Oracle Recruiting Cloud plus Talent Acquisition Cloud, Greenhouse Recruiting plus Greenhouse Inclusion plus Greenhouse Connect, Lever ATS plus LeverTRM, iCIMS Talent Cloud, Personio Recruiting (DACH plus EU), SmartRecruiters, Workable, Recruitee, BambooHR ATS - with full per-applicant record access including application date, source, requirement-profile match, recruiter disposition, plus complete decision history
- AI-talent-intelligence integration: Eightfold AI Talent Intelligence, Beamery, HiringSolved, Hiretual, Pymetrics, HireVue Video Interviewing, Modern Hire (HireVue), Plum - subject to EU AI Act 2024 Annex III(4)(a) high-risk AI compliance plus NYC Local Law 144 AEDT bias-audit requirement plus Illinois HB 645 plus Maryland HB 1202 consent requirements plus EEOC Strategic Enforcement Plan AI scrutiny plus FTC Section 5 oversight
- EU AI Act conformity assessment documentation per Article 16 (Annex VI internal control or Annex VII third-party route), EU declaration of conformity per Article 47, CE marking per Article 48, registration in EU AI database per Article 71, plus deployer fundamental rights impact assessment per Article 27, plus Article 26(7) worker-representative information
- Bias monitoring framework per EEOC Uniform Guidelines on Employee Selection Procedures 29 CFR Part 1607 with Four-Fifths Rule plus standard-deviation analysis per Hazelwood plus statistical-significance testing (Fisher's Exact Test, Chi-Square, Z-test); plus NYC Local Law 144 annual independent third-party AEDT bias audit; plus EU AI Act Article 10 data quality plus Article 15 accuracy
- Decision logging infrastructure per EU AI Act Article 12 record-keeping plus Article 26(6) deployer logs (6 months minimum) plus OFCCP Internet Applicant Rule 41 CFR 60-1.12 recordkeeping (2 years federal contractor) plus UK Equality Act tribunal disclosure plus GDPR Article 22 challengeability plus retention schedule per OFCCP plus EEOC plus EU Member State
- Works council or worker representative consultation per EU AI Act Article 26(7) plus German BetrVG (Works Constitution Act) plus French CSE (Comité Social et Économique) plus Italian Statuto dei Lavoratori plus Netherlands COR plus EU Information and Consultation Directive 2002/14/EC; (UK: ICE Information and Consultation of Employees Regulations 2004 where 50+ employees)
- Data Protection Impact Assessment per GDPR Article 35 plus UK GDPR plus ICO Guidance on AI and Data Protection 2023 plus EDPB Guidelines on Article 22 automated decision-making plus CNIL Guidance plus BfDI Guidance plus Garante Privacy Guidance plus AEPD Guidance
- Human-in-the-loop workflow ensuring no Article 22 GDPR solely-automated decisions plus EU AI Act Article 14 human oversight by individuals with necessary competence plus training plus authority to interpret AI output, decide not to use, override, or intervene
- OFCCP Internet Applicant recordkeeping platform plus EEO-1 Component 1 plus VETS-4212 protected-veteran reporting plus NYC Local Law 144 annual audit posting plus state-level pay-data reporting (California SB 1162 plus Colorado plus Illinois plus Massachusetts plus New York plus Washington)
Infrastructure Contribution
What this assessment contains: 9 slides for your leadership team
Personalised with your numbers. Generated in 2 minutes directly in your browser. No upload, no login.
- 1
Title slide - Process name, decision points, automation potential
- 2
Executive summary - FTE freed, cost per transaction before/after, break-even date, cost of waiting
- 3
Current state - Transaction volume, error costs, growth scenario with FTE comparison
- 4
Solution architecture - Human - rules engine - AI agent with specific decision points
- 5
Governance - EU AI Act, works council, audit trail - with traffic light status
- 6
Risk analysis - 5 risks with likelihood, impact and mitigation
- 7
Roadmap - 3-phase plan with concrete calendar dates and Go/No-Go
- 8
Business case - 3-scenario comparison (do nothing/hire/automate) plus 3×3 sensitivity matrix
- 9
Discussion proposal - Concrete next steps with timeline and responsibilities
Includes: 3-scenario comparison
Do nothing vs. new hire vs. automation - with your salary level, your error rate and your growth plan. The one slide your CFO wants to see first.
Show calculation methodology
Hourly rate: Annual salary (your input) × 1.3 employer burden ÷ 1,720 annual work hours
Savings: Transactions × 12 × automation rate × minutes/transaction × hourly rate × economic factor
Quality ROI: Error reduction × transactions × 12 × EUR 260/error (APQC Open Standards Benchmarking)
FTE: Saved hours ÷ 1,720 annual work hours
Break-Even: Benchmark investment ÷ monthly combined savings (efficiency + quality)
New hire: Annual salary × 1.3 + EUR 12,000 recruiting per FTE
All data stays in your browser. Nothing is transmitted to any server.
Candidate Screening Agent - Title VII, EU AI Act Annex III, NYC AEDT | Gosign
Initial assessment for your leadership team
A thorough initial assessment in 2 minutes - with your numbers, your risk profile and industry benchmarks. No vendor logo, no sales pitch.
All data stays in your browser. Nothing is transmitted.
Related Pages
Agent Blueprint Available
A full blueprint for Candidate Screening Agent - Title VII, EU AI Act Annex III, NYC AEDT | Gosign is available with micro-decision decomposition, industry variants, and implementation details.
View BlueprintRelated Agents
Executive Recruiting Agent - SOX 404, Dodd-Frank 953(b), UK SM&CR | Gosign
Cross-jurisdictional executive recruiting plus SOX 404 plus Dodd-Frank Section 953(b) CEO Pay Ratio plus UK SM&CR plus UK Companies Act 2006 plus EU Pay Transparency 2023/970 plus EU AI Act Annex III HR-Recruitment High-Risk plus Title VII pay equity multivariate regression in one platform - integrated workflow across UK + EU + US for Audit Committee, Compensation Committee, Nomination Committee, General Counsel, CHRO.
Interview Scheduling Agent - Title VII, EU AI Act Annex III, NYC LL 144 | Gosign
Title VII compliant interview scheduling plus US California SB 1162 plus Colorado plus NYC Local Law 32 pay-range disclosure plus EU Pay Transparency Directive 2023/970 plus EU AI Act 2024/1689 Annex III(4)(a) HR-Recruitment High-Risk plus NYC Local Law 144 AEDT bias audit plus Mobley v. Workday precedent plus ADA Title I reasonable accommodation plus EEOC Forbidden Questions Filter plus DPIA Article 35 plus Microsoft Outlook plus Google Calendar plus Calendly plus GoodTime Hire plus Cronofy plus multi-calendar conflict resolution plus time-zone plus visa verification - one auditable scheduling pipeline across UK plus EU plus US for Talent Acquisition Director plus CHRO plus DPO plus Compliance Officer plus EEO Officer plus Internal Audit.
Frequently Asked Questions
How does the Agent handle EU AI Act 2024 Annex III(4)(a) High-Risk classification plus Article 16 conformity assessment plus Article 47 declaration plus Article 48 CE marking plus Article 26(7) worker-representative information plus Article 27 fundamental rights impact assessment for candidate-screening AI?
Candidate-screening AI is unambiguously classified as High-Risk under EU AI Act 2024 Article 6 plus Annex III(4)(a) covering AI systems intended to be used for the recruitment or selection of natural persons - in particular to place targeted job advertisements, to analyse and filter job applications, and to evaluate candidates. The deadline for compliance is 2 August 2026. The Agent operationalises EU AI Act compliance in five integrated phases. Phase 1 (Provider Conformity Assessment): assess whether the screening system is provider-developed (full provider obligations including Article 9 risk management plus Article 10 data quality plus Article 11 technical documentation plus Article 12 record-keeping plus Article 13 transparency plus Article 14 human oversight plus Article 15 accuracy plus Article 16 conformity assessment plus Article 47 declaration of conformity plus Article 48 CE marking) or deployer-customised (deployer obligations under Article 26 plus Article 27 FRIA). Phase 2 (Deployer Obligations): complete Article 27 fundamental rights impact assessment covering affected groups plus reasonably foreseeable risks plus mitigation measures plus human-oversight arrangements; inform worker representatives under Article 26(7) before introducing the system; ensure input data is relevant plus sufficiently representative under Article 26(4); operate per provider instructions plus monitor for serious incidents under Article 26(5); keep automatically-generated logs for at least 6 months under Article 26(6). Phase 3 (Risk Management): implement Article 9 risk management system covering identification plus analysis plus estimation plus evaluation plus elimination or mitigation of foreseeable risks throughout the lifecycle; integrate with EU AI Act Article 17 quality management system plus Article 79 corrective action plus Article 73 serious incident reporting (15-day deadline). Phase 4 (Transparency and Human Oversight): provide Article 13 transparency to deployers including instructions for use plus capabilities plus limitations plus level of accuracy plus risks; ensure Article 14 human oversight by individuals with the necessary competence plus training plus authority to interpret AI output, decide not to use AI output, override AI output, or intervene in AI operation. Phase 5 (Conformity and CE Marking): complete Article 16 conformity assessment under Annex VI internal control (default) or Annex VII third-party route (if no harmonised standards applicable); issue Article 47 EU declaration of conformity plus Article 48 CE marking; register in EU AI database under Article 71; plus Article 64-70 market surveillance cooperation.
How does the Agent process EEOC Four-Fifths Rule plus Adverse Impact Ratio plus disparate-impact analysis plus Title VII plus ADEA plus ADA plus GINA plus EEOC Uniform Guidelines on Employee Selection Procedures 29 CFR Part 1607?
Disparate-impact analysis is operationally complex because Title VII plus ADEA plus ADA plus GINA all create distinct protected-characteristic frameworks plus the EEOC Uniform Guidelines establish the Four-Fifths Rule (selection rate for any protected group less than 80% of highest-rate group) as the standard adverse-impact threshold under Griggs v Duke Power Co 1971 plus Hazelwood School District v United States 1977 standard-deviation framework. The Agent operationalises disparate-impact analysis in five integrated phases. Phase 1 (Protected-Group Identification): identify protected groups per EEOC Component 1 race/ethnicity categories (Hispanic or Latino, White, Black or African American, Native Hawaiian or Other Pacific Islander, Asian, American Indian or Alaska Native, Two or More Races) plus sex (M/F per EEOC) plus age 40+ plus disability self-identification plus protected-veteran status; collect demographic data through voluntary self-identification with separate channel from selection process to avoid GINA Section 202 prohibition. Phase 2 (Selection-Rate Calculation): calculate selection rate per protected group as (number selected / number considered) per EEOC definition plus OFCCP Internet Applicant Rule definition (submitted expression of interest plus considered for position plus possessed basic qualifications plus did not self-remove); compute Impact Ratio as (selection rate of group / selection rate of highest-rate group). Phase 3 (Four-Fifths Rule Test): apply Four-Fifths Rule threshold (Impact Ratio less than 0.80 indicates adverse impact requiring business-necessity justification under Griggs v Duke Power Co); supplement with statistical-significance testing (Fisher's Exact Test, Chi-Square, Z-test for two proportions) plus standard-deviation analysis per Hazelwood (greater than 2-3 standard deviations indicates statistical significance). Phase 4 (Business-Necessity Justification): if adverse impact identified, document business-necessity per Title VII Section 703(k) burden-shifting framework requiring (a) job-related to position plus (b) consistent with business necessity plus (c) consideration of less-discriminatory alternatives; integrate with Uniform Guidelines validation evidence (criterion-related, content-related, construct-related). Phase 5 (Remediation and Documentation): if business necessity not established or less-discriminatory alternative available, modify selection procedure; document analysis plus remediation per OFCCP Internet Applicant Rule 2-year recordkeeping plus EEOC EEO-1 reporting plus integration with NYC Local Law 144 AEDT annual independent third-party audit.
How does the Agent handle US OFCCP Internet Applicant Rule 41 CFR 60-1.12 plus Executive Order 11246 plus Section 503 plus VEVRAA federal contractor recordkeeping plus affirmative-action programs?
Federal contractor obligations under OFCCP create unique recordkeeping requirements beyond EEOC standards because Executive Order 11246 plus Section 503 plus VEVRAA require affirmative-action programs plus protected-veteran plus disability self-identification plus Internet Applicant recordkeeping for every individual meeting four-part test. The Agent operationalises OFCCP compliance in five integrated phases. Phase 1 (Federal Contractor Status): determine OFCCP coverage based on federal contract value (USD 50,000+ Executive Order 11246 plus Section 503; USD 150,000+ VEVRAA), establish coverage threshold per contracting agency plus duration plus modification; integrate with System for Award Management (SAM.gov) plus Federal Procurement Data System (FPDS) plus federal-contract recordkeeping. Phase 2 (Internet Applicant Identification): apply 41 CFR 60-1.12 four-part test for each individual: (a) submitted expression of interest through internet/related electronic technologies, (b) considered for the particular position, (c) possessed basic qualifications, (d) did not at any point in the selection process remove self from consideration; record per-applicant data including race/ethnicity self-identification plus sex plus protected-veteran self-identification plus disability self-identification (7% utilisation goal). Phase 3 (Affirmative-Action Program Documentation): maintain affirmative-action program covering organisational profile plus job-group analysis plus availability analysis plus utilisation analysis plus goals plus action-oriented programs; integrate with OFCCP Functional Affirmative Action Programs (FAAP) plus Corporate Management Compliance Evaluation plus desk audits plus on-site reviews. Phase 4 (Pre-Offer Self-Identification): collect protected-veteran self-identification under VEVRAA plus disability self-identification under Section 503 with required language plus voluntary disclosure plus separate document plus retention separate from personnel file; comply with EEOC ADA pre-offer disability inquiry restriction plus GINA genetic-information prohibition. Phase 5 (Audit and Reporting): retain Internet Applicant data 2 years plus respond to OFCCP audit request including itemised listing plus race/ethnicity plus sex plus protected-veteran plus disability per applicant; integrate with VETS-4212 protected-veteran annual filing plus EEO-1 Component 1 plus Component 2 reporting; plus Predetermination Notice plus Notice of Violation plus Conciliation Agreement workflow.
How does the Agent comply with NYC Local Law 144 of 2021 AEDT bias-audit requirement plus Illinois HB 645 Artificial Intelligence Video Interview Act plus Maryland HB 1202 facial recognition plus California SB 1162 pay-disclosure?
US state-level AI hiring laws create overlapping compliance requirements because NYC Local Law 144 of 2021 (effective 5 July 2023) was the first jurisdictional bias-audit mandate, followed by Illinois HB 645 (2020) plus Maryland HB 1202 (2020) plus California SB 1162 (2022) plus Colorado AI Act (2024). The Agent operationalises US state-level AI hiring compliance in five integrated phases. Phase 1 (NYC Local Law 144 AEDT Identification): identify whether the screening tool qualifies as Automated Employment Decision Tool (AEDT) under NYC Department of Consumer and Worker Protection rule covering computational processes derived from machine learning, statistical modeling, data analytics, or AI that issue simplified output, including a score, classification, or recommendation, used to substantially assist or replace discretionary decision making for employment decisions affecting candidates plus employees in NYC; coverage extends to remote roles where employee resides in NYC. Phase 2 (NYC Annual Bias Audit): commission independent third-party auditor (separate from AEDT vendor plus from employer) to conduct annual bias audit covering selection rate per sex, race/ethnicity, intersectional EEOC Component 1 categories plus impact ratio plus statistical methodology; audit must use historical data of past 1 year (where available) or test data; audit summary must be posted on employer website with clear notice of date plus group categories plus selection rates plus impact ratios. Phase 3 (NYC Candidate Notice): provide 10-business-day notice to candidates plus employees before AEDT use including notice of AEDT use, characteristics evaluated, alternative selection process or accommodation request mechanism; integrate with EEOC ADA reasonable-accommodation framework. Phase 4 (Illinois HB 645 plus Maryland HB 1202): obtain candidate consent before AI video interview under Illinois HB 645 with notice of AI evaluation plus characteristics evaluated plus video destruction within 30 days; obtain candidate consent before facial recognition under Maryland HB 1202; integrate with HireVue Video Interviewing plus Modern Hire (HireVue) plus consent-management workflow. Phase 5 (California SB 1162 plus Colorado AI Act): comply with California SB 1162 pay-scale disclosure in job postings (15+ employees) plus pay-data reporting (100+ employees) plus retention requirements; comply with Colorado AI Act 2024 high-risk AI deployer obligations including impact assessment plus consumer notification plus reasonable care to avoid algorithmic discrimination; integrate with state-level enforcement coordination.
How does the Agent process UK Equality Act 2010 plus EHRC Code of Practice plus Section 60 prohibition on pre-employment health enquiries plus reasonable-adjustment duty?
UK Equality Act 2010 creates the most consolidated discrimination framework covering nine protected characteristics with distinct rules for direct discrimination plus indirect discrimination plus harassment plus victimisation plus reasonable-adjustment duty plus public-sector equality duty. The Agent operationalises UK Equality Act compliance in five integrated phases. Phase 1 (Protected Characteristic Identification): identify nine protected characteristics under Section 4 (age, disability, gender reassignment, marriage and civil partnership, pregnancy and maternity, race, religion or belief, sex, sexual orientation); structure data collection plus monitoring plus reporting per ICO Employment Practices Code plus EHRC Statutory Code of Practice on Employment plus Section 159 positive action provisions. Phase 2 (Section 60 Pre-Employment Health Enquiries): apply Section 60 prohibition on pre-employment health-related enquiries except for narrow exceptions including (a) intrinsic-function inquiries (whether candidate can carry out function intrinsic to work with reasonable adjustments), (b) monitoring diversity, (c) positive action for disabled persons, (d) ensuring access to recruitment process, (e) determining genuine occupational requirement; integrate with reasonable-adjustment workflow under Section 20 plus Schedule 8. Phase 3 (Indirect Discrimination Test): assess each requirement criterion against Section 19 indirect discrimination test - provision/criterion/practice (PCP) applied to all, particular disadvantage to protected group, individual disadvantage, plus objective justification (proportionate means of achieving legitimate aim); document business-necessity rationale plus less-discriminatory alternatives consideration. Phase 4 (Reasonable Adjustment Duty): operationalise reasonable-adjustment duty under Section 20 plus Schedule 8 covering provisions/criteria/practices, physical features, auxiliary aids; integrate with disability self-identification plus interactive process plus reasonable-cost analysis; comply with Equality Act 2010 (Disability) Regulations 2010. Phase 5 (Tribunal-Defence Documentation): maintain documentation supporting tribunal defence including structured-interview consistency plus criterion application plus rationale plus reasonable-adjustment offer plus less-discriminatory alternative consideration; integrate with EHRC formal investigation plus unlawful act notice workflow plus tribunal disclosure requirements.
How does the Agent integrate with Workday Recruiting, SAP SuccessFactors Recruiting, Oracle Recruiting Cloud, Greenhouse, Lever, iCIMS, Eightfold AI, Pymetrics, HireVue, and Personio Recruiting?
The candidate-screening landscape spans the HCM-embedded ATS layer plus the dedicated ATS layer plus the AI-talent-intelligence layer plus the video-interviewing layer plus the assessment layer - and the Agent operates as the integration point across all five with EU AI Act compliance gating. HCM-embedded ATS: Workday Recruiting plus Workday Talent Optimization plus Workday Skills Cloud provide cloud-native recruiting embedded in Workday HCM with structured candidate profile plus competency-based screening plus internal-mobility plus referral plus campus-recruiting workflows; SAP SuccessFactors Recruiting plus Recruiting Marketing plus Onboarding provide enterprise recruiting suite with 50+ country localisation tightly integrated with SAP S/4HANA HR; Oracle Recruiting Cloud plus Talent Acquisition Cloud provide enterprise recruiting integrated with Oracle Fusion Cloud HCM plus Performance Management plus Compensation; ADP Recruiting Management provides recruiting integrated with ADP Workforce Now payroll. Dedicated ATS: Greenhouse Recruiting plus Greenhouse Inclusion provide market-leading dedicated ATS for tech plus high-growth companies plus mid-market with structured-interview kits plus scorecard methodology plus blind initial screening; Lever ATS plus LeverTRM provide ATS plus CRM combining recruiting with passive-candidate sourcing plus pipeline nurturing; iCIMS Talent Cloud provides enterprise ATS with 700+ marketplace partners particularly strong in regulated industries requiring OFCCP Internet Applicant Rule recordkeeping; SmartRecruiters plus Workable plus Recruitee plus JazzHR plus BreezyHR plus Bullhorn ATS plus Avature plus Lattice Recruiting cover SMB plus mid-market plus staffing-agency segments; Personio Recruiting dominates DACH plus EU mid-market with UK plus EU Member State localisation. AI talent-intelligence: Eightfold AI Talent Intelligence plus Beamery plus HiringSolved plus Hiretual provide AI-driven skills-graph-based matching plus internal-mobility recommendations plus diversity sourcing - all subject to EU AI Act 2024 Annex III(4)(a) high-risk AI compliance plus NYC Local Law 144 AEDT bias-audit requirement plus EEOC Strategic Enforcement Plan AI scrutiny. Video-interviewing plus assessment: HireVue Video Interviewing plus Modern Hire (now HireVue) plus Pymetrics plus Plum provide video-interview assessment plus game-based assessment plus personality assessment subject to Illinois HB 645 plus Maryland HB 1202 consent requirements plus EU AI Act compliance plus EEOC ADA reasonable-accommodation framework. Executive search plus RPO: Korn Ferry plus Heidrick & Struggles plus Russell Reynolds plus Mercer plus Aon plus Willis Towers Watson provide senior-level plus board-level plus C-suite recruiting plus assessment-validation plus structured-interview-design. The Agent operates as the upstream EU AI Act conformity-assessment plus bias-monitoring plus Article 14 human-oversight layer feeding the downstream ATS workflow, or the orchestration layer running parallel deployments where different business units use different ATS systems post-acquisition.
What Happens Next?
30 minutes
Initial call
We analyse your process and identify the optimal starting point.
1 week
Discover
Mapping your decision logic. Rule sets documented, Decision Layer designed.
3-4 weeks
Build
Production agent in your infrastructure. Governance, audit trail, cert-ready from day 1.
12-18 months
Self-sufficient
Full access to source code, prompts and rule versions. No vendor lock-in.
Implement This Agent?
We assess your process landscape and show how this agent fits into your infrastructure.