Compliance

EU AI Act Compliance Statement

How GetPruf meets the transparency, human-oversight, and risk-management requirements for high-risk AI systems in employment use cases.
Effective: August 2, 2026

Classification

GetPruf is classified as a high-risk AI system under the EU AI Act, Article 6(2), in conjunction with Annex III, point 4 (employment, workers management and access to self-employment). AI systems used to screen or filter applications, or to evaluate candidates in recruitment processes, fall within this category.
We are committed to meeting all requirements for high-risk AI systems by the August 2, 2026 compliance deadline. This page documents our compliance measures.

Transparency

GetPruf provides clear, accessible documentation of what the AI system does and does not do:
  • The AI extracts structured facts from resume text (employment dates, skills, claims, metrics)
  • Deterministic Python code computes all scores - the AI does not assign risk scores directly
  • Six weighted dimensions are scored independently, each with documented weights and methodology
  • Web verification checks entity existence only - it does not confirm employment or enrollment
  • The system does not access private databases, social media accounts, or protected characteristics
Full methodology documentation is available at /methodology. Every PDF report includes a "How This Score Was Calculated" section showing the exact contribution of each factor.

Human Oversight

GetPruf is designed as a decision-support tool, not a decision-making tool. Human oversight requirements are built into every stage:
No automated decisionsThe system never makes hiring decisions, rejects candidates, or takes adverse actions automatically.
Mandatory human reviewEvery report includes a clear statement that human review is required before any hiring decision.
Explanation of resultsReports provide specific concerns and suggested interview questions, not pass/fail verdicts.
Override capabilityHR teams can dismiss any flagged concern based on their professional judgment and candidate communication.

Bias Mitigation

GetPruf implements multiple safeguards to detect and reduce discriminatory outcomes:
  • Non-native English speaker adjustment: Language Quality and Authorship Signals scoring reduced by up to 40% when non-native writing patterns are detected
  • Industry-aware scoring: Metadata and web verification weights reduced by 40% for industries where companies typically have limited web presence
  • Career length calibration: Template-like language scored less harshly for early-career candidates
  • Four-fifths rule monitoring: Automated bias audit applies the EEOC four-fifths rule to detect disparate impact across candidate segments
  • No protected characteristics: The system does not process or consider race, gender, age, religion, disability, or national origin

Risk Management

Technical risk management measures include:
  • Deterministic scoring: AI extracts facts, code computes scores. The same resume always produces the same score.
  • Capped adjustments: Bayesian verification adjustment is capped at +40/-10 points to prevent single-factor dominance
  • Inconclusive default: Absence of web evidence is treated as inconclusive, not as evidence of fabrication
  • Prompt injection protection: Resume text is sanitized before AI processing to prevent adversarial inputs
  • Fail-closed security: When any component fails (rate limiter, auth, quota), the default is deny, not allow

Data Governance

Resume data is processed in accordance with our Privacy Policy and Data Processing Agreement. Key measures:
  • Multi-tenant isolation with per-query tenant_id scoping
  • TLS 1.3 encryption in transit, AES-256 at rest
  • Resume files processed in memory and deleted after analysis
  • No use of personal data for AI model training
  • Sub-processors documented with data handling practices reviewed

Technical Documentation

The following documentation is publicly available:
Additional technical documentation, including detailed model cards and risk assessments, is available on request for regulatory authorities and enterprise customers.

Contact

For AI Act compliance inquiries: dpo@getpruf.ai