AI Ethics Principles & Guidelines

Version: 2.0   Effective: June 2026   Classification: Public 

Supersedes: Version 1.0   •   Entity: NetConnect Private Limited, operating as Welocity™ (CIN U32202KA1997PTC021881), Bangalore, India 

Our Commitment to Ethical AI 

At Welocity.ai, we recognize the profound impact that AI-assisted recruitment technology has on individuals’ careers, organizations’ success, and society’s progress toward equitable employment. This responsibility guides how we design, develop, deploy, and continuously improve our AI-assisted interview and hiring platform. 


These principles reflect our commitment to responsible AI. They are living guidelines that evolve with technological advancement, regulatory change, and societal expectations. We engage with customers, candidates, ethicists, legal experts, industrial-organizational (IO) psychologists, and the wider community to maintain high ethical standards. 

Core AI Ethical Principles 

1. Human-Centered Design 


We augment human decision-making and protect human dignity. 


  • Our AI systems are designed to support, not replace, human judgment in hiring decisions. 


  • We maintain meaningful human oversight at every critical decision point. 


  • We treat candidates with respect through transparent, accessible interview experiences. 


  • We give recruiters useful insights while preserving their ultimate decision-making authority. 


2. Fairness and Non-Discrimination 


We work to prevent unlawful bias and to support diversity and equal opportunity. 


  • We apply bias-detection and mitigation techniques across the AI lifecycle. 


  • We design and test our models to avoid producing unlawful discrimination based on protected characteristics such as race, color, gender, age, disability, religion, sexual orientation, or other status protected by applicable law. 


  • We regularly test for adverse impact across demographic groups and act on what those tests show. 


  • We design our assessments to support equal opportunity in employment. 



3. Transparency and Explainability 


We help candidates and customers understand how our AI is used. 


  • We clearly disclose when and how AI is used in the assessment process. 


  • We provide results that recruiters can understand and reasonably act upon. 


  • We give candidates accessible information about the assessment process and the criteria assessed. 


  • We maintain documentation of our AI models, their intended purpose, and their decision logic. 


4. Privacy and Data Protection 


We safeguard personal data and apply privacy-by-design. 


  • We apply privacy-by-design and data-minimization principles in our AI systems. 


  • We collect only the data necessary for legitimate, job-related assessment. 


  • We provide clear consent mechanisms and support data-subject rights. 


  • We apply appropriate security to sensitive data, including any audio or video recordings and, where processed, special-category or biometric data. 


  • We design our practices to align with applicable data-protection laws, including the GDPR, the UK GDPR, and India’s Digital Personal Data Protection Act, 2023, and, where relevant to our operations, laws such as the California CCPA/CPRA and the Illinois BIPA. 


5. Accountability and Governance 


We take responsibility for the impact of our AI systems. 


  • We maintain clear governance structures for AI development and deployment. 


  • We define accountability for AI-related decisions. 


  • We provide channels for feedback, concerns, and redress. 


  • We conduct periodic ethical reviews of our AI practices. 


6. Scientific Validity and Reliability 


We keep our assessments scientifically grounded. 


  • We base our algorithms on established industrial-organizational psychology principles. 


  • We validate our assessments against relevant, job-related performance criteria. 


  • We test for reliability and consistency in our measurements. 


  • We work with IO psychologists and data scientists to maintain scientific rigor. 

Bias Prevention and Mitigation Framework

We apply a multi-layered approach across the AI lifecycle: 

Pre-Development Analysis 


  • Diverse and representative training-data collection. 


  • Stakeholder consultation including diverse perspectives. 


  • Ethical and impact assessment for new AI features. 


During Development 


  • Algorithmic fairness constraints incorporated into model training. 


  • Regular bias testing across protected characteristics. 


  • Feature selection that excludes bias-inducing variables. 


  • Cross-functional review by diverse teams. 


Pre-Deployment Testing 


  • Comprehensive adverse-impact analysis. 


  • Validation aligned with the EEOC Uniform Guidelines, where US roles are in scope. 


  • Third-party audits where applicable. 



  • Pilot testing with diverse candidate populations. 


Post-Deployment Monitoring 


  • Continuous monitoring of model performance across demographics. 


  • Regular fairness audits and reporting. 


  • Feedback loops for improvement. 


  • Rapid-response protocols for identified issues. 

Compliance with Legal Standards 

We design our practices to align with, and where applicable comply with: 


  • The EU AI Act — recruitment and candidate-evaluation systems are classified as high-risk under Annex III. We work to meet the corresponding obligations (risk management, data governance, transparency, human oversight, record-keeping, and accuracy/robustness), and we do not deploy practices the Act prohibits, including emotion recognition in the employment context and untargeted facial-image scraping. 


  • The GDPR and UK GDPR, and India’s Digital Personal Data Protection Act, 2023. 


  • The EEOC Uniform Guidelines on Employee Selection Procedures (1978), where US roles are in scope. 


  • State and local automated-employment-decision laws, such as NYC Local Law 144, where applicable. 


  • International standards including ISO/IEC 23053:2022 and ISO/IEC 23894:2023. 

AI Model Development Process 

Phase 1: Job Analysis and Design 


  • Comprehensive job analysis to identify relevant competencies. 


  • Define clear, measurable performance indicators. 


  • Design structured interview questions based on IO psychology research. 


  • Establish validation criteria for model success. 


Phase 2: Data Collection and Preparation 



  • Collect diverse, representative training data. 


  • Implement data-quality controls. 


  • Apply privacy-preserving techniques. 


  • Create balanced datasets across demographic groups. 


Phase 3: Model Development 


  • Train models using current natural-language-processing techniques on interview audio and transcripts. 


  • Focus on job-relevant features (e.g., communication content) rather than appearance. 


  • Incorporate fairness constraints during training. 


  • Favor explainable model architectures. 


Phase 4: Bias Testing and Mitigation 


  • Conduct bias audits. 


  • Analyze adverse impact across protected groups. 


  • Remove or adjust bias-inducing features. 


  • Re-train models with fairness optimization and validate improvements. 


Phase 5: Validation and Deployment 


  • Validate predictive validity against job-related performance. 


  • Test reliability across different contexts. 


  • Conduct final fairness assessments. 


  • Deploy with monitoring systems in place. 


Phase 6: Continuous Improvement 



  • Monitor real-world performance. 


  • Collect feedback from users and candidates. 


  • Re-train periodically with new data. 


  • Conduct periodic third-party audits and update models as job requirements change. 

Specific AI Technologies and Their Ethical Safeguards 

Natural Language Processing (NLP) 


  • What we analyze: the content, structure, and relevance of responses. 


  • What we avoid: penalizing accent, dialect, or speech patterns that may correlate with protected characteristics. 


  • Safeguards: dialect-neutral processing and review for language bias. 


Audio and Transcript Analysis 


  • What we analyze: the audio and transcript of the interview — language fluency, relevance and coherence of responses, technical accuracy where applicable, and communication clarity. 


  • What we avoid: penalizing accent, dialect, or speech patterns that may correlate with protected characteristics. 


  • Safeguards: dialect-neutral processing and review for language bias. 


Video Recording (not analyzed by AI) 


  • What we do: we capture and store the interview video so that human reviewers — Welocity recruiters and authorized client recipients — can view it as part of their own evaluation. 


  • What we do not do: we do not apply computer-vision analysis to the video. We do not perform emotion recognition or affective-state inference in the employment context (a practice prohibited under the EU AI Act), nor facial-recognition identification or biometric categorization for assessment purposes. 


  • Safeguards: the video informs human judgment only; AI evaluation is based on audio and transcript, not visual content. 


Behavioral Assessment 


  • What we measure: job-relevant competencies and skills. 


  • What we avoid: personality inferences unrelated to job performance. 


  • Safeguards: competency-based frameworks validated against job outcomes. 

Candidate Rights and Protections 

We support the following candidate rights: 


  • Information about AI use in their assessment. 


  • Understanding of the assessment criteria and process. 


  • Reasonable accommodation for disabilities or special needs. 


  • Access to their personal data and assessment results, where legally required. 


  • Correction of inaccurate personal information. 


  • Human review of AI-based decisions, where applicable. 


  • The ability to opt out of certain AI processing, subject to employer policies and applicable law. 


  • The ability to file complaints about AI assessment practices. 

Governance and Oversight

AI Ethics Committee 


  • Periodic reviews (at least quarterly) of AI practices and outcomes. 


  • Investigation of ethical concerns. 


  • Guidance on emerging ethical challenges. 


  • Stakeholder engagement and consultation. 


Representative Composition (by role) 


  • Chief AI Officer 


  • Head of Engineering / Data Science 



  • Industrial-Organizational Psychologist(s) 


  • Legal & Compliance 


  • Diversity, Equity & Inclusion representative(s) 


  • External ethics advisor(s) 


Continuous Education 


  • Regular AI-ethics training for relevant team members. 


  • Participation in industry forums and standards bodies. 


  • Collaboration with academic researchers. 


  • Engagement with regulatory bodies. 

Measurement and Reporting 

Key Metrics We Track

Fairness Metrics 


  • Demographic parity across groups. 


  • Equalized odds and equal opportunity. 


  • Adverse-impact ratios. 


Performance Metrics 


  • Predictive-validity coefficients. 


  • False positive / negative rates by group. 


  • Model accuracy and reliability. 


Transparency Metrics 


  • Explainability scores. 


  • User-understanding assessments. 


  • Candidate-satisfaction ratings. 


Regular Reporting 


  • Annual AI Ethics Report (public). 


  • Quarterly internal ethics reviews. 


  • Customer-specific bias-audit reports, on request. 


  • Regulatory compliance documentation. 

Commitment to Continuous Improvement 

Ethical AI is an ongoing journey, not a destination. We commit to: 


  • Staying current with evolving ethical standards and best practices. 


  • Listening actively to feedback from all stakeholders. 


  • Investing in bias-mitigation research and development. 


  • Collaborating openly with the broader AI-ethics community. 


  • Adapting to new challenges and opportunities. 


  • Leading responsibly within the recruitment-technology industry. 

Contact and Feedback 

We welcome dialogue about our AI-ethics practices: 


Email: ethics@welocity.ai 


Website: https://welocity.ai/ai-ethics 


Ethics Hotline (anonymous): https://welocity.ai/ethics-concerns 


Data protection: privacy@welocity.ai 


Data Protection Officer: dpo@welocity.ai 


EU Representative (GDPR Article 27): an EU-based representative is being appointed as the contact point for individuals in the European Economic Area and supervisory authorities; current details are published at www.welocity.ai/legal. 


Registered entity: NetConnect Private Limited, operating as Welocity™ — CIN U32202KA1997PTC021881 — Registered office: WeWork Salarpuria Symbiosis, Bannerghatta Main Road, Arekere, Bengaluru, Karnataka 560076, India. 

References and Standards 

Our AI-ethics framework is informed by: 


  • IEEE Standards for Ethical AI (P7000 series). 


  • ISO/IEC 23053:2022 — Framework for AI systems using machine learning. 


  • ISO/IEC 23894:2023 — AI risk management. 


  • Partnership on AI — Tenets and best practices. 


  • OECD AI Principles (2019). 


  • EU Ethics Guidelines for Trustworthy AI. 


  • Asilomar AI Principles. 


  • Montreal Declaration for Responsible AI. 


  • ACM Code of Ethics and Professional Conduct. 


Last Updated: June 2026  •  Next Review: Quarterly  •  Classification: Public 

© 2026 NetConnect Private Limited. All rights reserved.