shortlistd.io AI Policy | Responsible AI in Recruiting
Last Updated: December 2, 2025
Purpose and Commitment
Shortlistd is committed to building and deploying artificial intelligence responsibly, ethically, and in compliance with emerging AI regulations worldwide. This policy establishes our framework for developing AI-powered recruitment tools that are fair, transparent, accountable, and respect human dignity and rights.
This policy applies to all AI systems, models, and algorithms used in our Services, including:
Candidate discovery and sourcing
Contact information enrichment
Resume analysis and skill extraction
Interview assistance and evaluation
Candidate-job matching and recommendations
Automated communication and outreach
Our Core Principles:
Human Dignity: AI serves humans; humans always make final decisions
Fairness: Actively prevent bias and discrimination
Transparency: Clear explanations of how AI works and why it makes recommendations
Accountability: Clear responsibility for AI outcomes
Privacy: Minimize data collection and respect individual rights
Safety: Prevent harm and misuse
1. Human-in-the-Loop: AI as Assistant, Not Decision-Maker
1.1 Mandatory Human Oversight
Absolute Rule: Our AI systems NEVER make final hiring, rejection, or employment decisions without human review and approval.
What This Means in Practice:
AI generates recommendations, suggestions, and insights
Humans (recruiters, hiring managers) review AI output
Humans make all final decisions about candidates
Humans can override, ignore, or modify any AI recommendation
Humans remain accountable for all hiring outcomes
Prohibited: Automated rejection, automated ranking that excludes candidates from consideration, or any system where AI makes binding decisions without human intervention.
1.2 Levels of AI Involvement
Level 1 - AI Assistance Only (Our Current Approach):
AI discovers and surfaces relevant candidates
AI provides information summaries and skill extraction
AI suggests questions or evaluation criteria
Human makes all selection and contact decisions
This is our standard operating model
Level 2 - AI Scoring with Human Review:
AI generates competency scores or match percentages
Scores are advisory and accompanied by explanations
Humans review scores and underlying reasoning
Humans make final decisions
Candidates can request manual re-evaluation
Prohibited - Level 3 - Fully Automated Decisions:
AI automatically rejects candidates → NOT ALLOWED
AI automatically advances candidates without review → NOT ALLOWED
AI makes binding decisions → NOT ALLOWED
1.3 Candidate Rights Regarding AI
Right to Human Review:
Any candidate can request that a human review AI-generated assessments
No additional cost or penalty for requesting human review
Responses provided within reasonable timeframes
Right to Explanation:
Candidates can ask how AI evaluated their profile
We provide clear, understandable explanations
We identify which factors influenced AI recommendations
Right to Challenge:
Candidates can contest AI-generated information
We provide mechanisms to correct errors
We escalate disputes to human reviewers
2. Bias Prevention and Fairness
2.1 Prohibited: Sensitive Attribute Inference and Use
Absolute Prohibition: Our AI systems are designed to NOT infer, generate, process, or use the following sensitive characteristics in any way:
Protected Categories We Do NOT Process:
Race or ethnicity
National origin or citizenship status (except where legally required for work authorization)
Gender identity or sex
Sexual orientation
Religious beliefs or philosophical views
Political opinions or affiliations
Trade union membership
Health information, medical conditions, or disability status (except reasonable accommodations voluntarily disclosed)
Genetic data or biometric identifiers
Pregnancy or family planning
Age (except minimum legal working age verification)
Marital or family status
Technical Implementation:
Input filters to remove sensitive data from processing
Output sanitization to prevent generation of sensitive inferences
Regular auditing of AI outputs for accidental bias indicators
Training data curation to exclude discriminatory patterns
If Sensitive Data Appears:
Clients must not use such information in hiring decisions
Clients must report it immediately to info@shortlistd.io
We will investigate and improve filtering
Affected candidates will be notified if their data was inappropriately processed
2.2 Fairness-by-Design Approach
Principle: Our AI systems are designed from the ground up to minimize bias and promote fair evaluation.
Design Choices:
Competency-Based Evaluation:
Focus on demonstrated skills and relevant experience
Avoid proxies that correlate with protected characteristics
Use standardized, job-relevant criteria
Structured Assessment:
Standardized questions and evaluation frameworks
Consistent criteria applied to all candidates
Reduces subjective human bias
Diverse Training Data:
Training data represents diverse candidate populations
Actively balance datasets to prevent demographic skews
Regular data quality audits
Fairness Testing:
Regular testing for disparate impact across demographic groups
Statistical analysis of hiring outcomes
Third-party bias audits (planned as we scale)
2.3 Bias Monitoring and Mitigation
Ongoing Monitoring:
Track hiring outcomes by client and role type
Analyze patterns that might indicate bias
Collect and review candidate feedback
Monitor AI recommendations for potential discriminatory patterns
Corrective Actions:
Immediate investigation of bias indicators
Model retraining or adjustment when bias detected
Client notification and guidance on fair practices
Continuous improvement of fairness safeguards
Transparency in Limitations:
We acknowledge that no AI system is perfect
We openly communicate known limitations
We commit to continuous improvement
We invite feedback and external scrutiny
3. Transparency and Explainability
3.1 How Our AI Works
High-Level Process:
Step 1: Candidate Discovery
AI searches public sources for candidates matching job requirements
Uses semantic search and natural language understanding
Retrieves publicly available professional information
Step 2: Data Enrichment
AI extracts and structures information from profiles
Identifies skills, experience levels, and qualifications
Finds contact information from legitimate public sources
Step 3: Matching and Ranking
AI compares candidate profiles to job requirements
Generates match scores based on relevant criteria
Provides explanations for match scores
Step 4: Assessment Assistance
AI helps structure interviews and evaluations
Suggests competency-based questions
Analyzes interview responses for key skills and qualifications
Generates summaries and evaluation support
Step 5: Human Decision
Human reviewers examine AI recommendations
Humans consider AI inputs alongside their own judgment
Humans make all final hiring decisions
3.2 Explainability Standards
For Every AI Recommendation, We Provide:
What: Clear statement of the recommendation or score
Why: Specific factors that influenced the recommendation
Evidence: Concrete examples from candidate data supporting the recommendation
Confidence: Indication of AI's certainty level
Limitations: What the AI did not consider or could not evaluate
Example Explanation:
"Match Score: 85% Reasoning: Candidate has 5+ years of experience with Python and demonstrated expertise in machine learning projects (see GitHub portfolio). Relevant skills include TensorFlow and PyTorch, which match job requirements. Strong communication skills evidenced by technical blog posts. Limitations: AI cannot evaluate culture fit or soft skills not demonstrated in written content."
3.3 Transparency to Candidates
What Candidates Can Learn:
That AI was used in their evaluation
What information the AI analyzed
What factors influenced their match score or recommendation
How to request human review or contest AI outputs
How to Request Information:
Email info@shortlistd.io with "AI Processing Inquiry"
Include your name and relevant job application details
We respond within 30 days with clear, non-technical explanations
4. Data Minimization and Purpose Limitation
4.1 Collect Only What's Necessary
Principle: We collect and process only the personal data necessary for legitimate recruitment purposes.
What We Collect:
Professional work history and experience
Relevant skills and qualifications
Education and certifications
Public contact information (business email/phone)
Public portfolios and professional content
What We Do NOT Collect (Unless Voluntarily Provided):
Personal (non-business) contact information
Social security numbers or government IDs
Financial information (except for payroll, which we don't handle)
Personal social media content unrelated to professional qualifications
Private communications or documents
Sensitive personal information (see Section 2.1)
4.2 Purpose Limitation
Permitted Uses of Candidate Data:
Matching candidates with job opportunities
Facilitating communication between candidates and employers
Evaluating candidate qualifications for specific roles
Providing recruitment insights to clients
Improving our Services (using aggregated, anonymized data)
Prohibited Uses:
Marketing non-recruitment services to candidates without consent
Building or selling candidate databases
Training AI models on personal data without consent
Any use unrelated to recruitment and hiring
4.3 Storage and Retention Minimization
Data Lifecycle:
Collection: Gather only necessary data
Active Use: Process for specific recruitment purpose
Retention: Keep only as long as needed
Deletion: Remove when purpose is fulfilled or upon request
Retention Periods:
Active recruitment data: Duration of job search or client engagement
Cached search results: Maximum 90 days
After deletion request: Immediate removal + suppression list entry
See Privacy Policy Section 4 for detailed schedules
Suppression Lists:
Candidates who request deletion are permanently suppressed
Prevents re-discovery and re-processing
Honors "do not process" requests indefinitely
5. Consent and Control
5.1 Candidate Consent and Notification
For Publicly Available Data:
We rely on clients' legitimate interest in recruitment (GDPR lawful basis)
Candidates are notified when contacted for opportunities
Candidates can opt out at any time
Article 14 Notices (GDPR): When we collect candidate data indirectly (from public sources), we ensure:
Candidates are informed about our processing
Information about data sources, purposes, and retention
Clear instructions for exercising privacy rights
Notification within 30 days of collection or upon first contact
Explicit Consent Required For:
Processing beyond recruitment purposes
Training AI models on individual candidate data
Sharing data with third parties beyond recruitment clients
Any use of sensitive personal information
5.2 Client Consent and Requirements
Clients Using Our Services Must:
Establish lawful basis for processing candidate data
Provide required notices to candidates (or authorize us to do so)
Use candidate data only for recruitment purposes
Honor candidate privacy rights and deletion requests
Document consent where required
Sign our Data Processing Agreement
Client Responsibilities:
Ensure recruitment practices comply with employment laws
Avoid discrimination and bias in hiring decisions
Conduct due diligence beyond AI recommendations
Maintain records of hiring decisions and rationale
5.3 Individual Control and Rights
Candidates Can:
Request access to their data
Correct inaccurate information
Request deletion (with suppression to prevent re-discovery)
Object to processing
Restrict how their data is used
Request human review of AI assessments
Opt out of AI-assisted processing where legally required
How to Exercise Rights:
Email info@shortlistd.io with "Candidate Privacy Request"
Include your name and any known contact information
We respond within 30 days (or as required by applicable law)
6. AI Training Data and Model Governance
6.1 Training Data Standards
Data Sources for AI Training:
Publicly available, anonymized datasets
Synthetic and simulated data
Properly licensed commercial datasets
Aggregated, de-identified usage data (with consent)
Prohibited Training Data:
Personal data without explicit consent
Datasets known to contain bias or discriminatory patterns
Illegally obtained or scraped data
Proprietary data from competitors
Sensitive personal information
Data Quality and Curation:
Regular audits of training data for quality and fairness
Removal of biased or problematic data
Balanced representation across demographics
Documentation of data sources and lineage
6.2 Model Development and Testing
Development Standards:
Fairness testing throughout development lifecycle
Validation on diverse test datasets
Performance metrics that include fairness indicators
Documentation of model architecture and decision logic
Pre-Deployment Testing:
Bias testing across protected characteristics
Accuracy and performance benchmarks
Edge case and adversarial testing
User acceptance testing with real recruiters
Post-Deployment Monitoring:
Ongoing performance monitoring
Real-world bias detection
Feedback collection from users and candidates
Regular model audits and updates
6.3 AI Model Transparency
Documentation We Maintain:
Model architecture and design decisions
Training data sources and characteristics
Known limitations and failure modes
Performance metrics and fairness evaluations
Version history and change logs
Available Upon Request:
General information about how our AI works
Explanation of specific recommendations
Information about data sources and processing
Details on fairness testing and bias mitigation
Proprietary Protection:
We balance transparency with protection of trade secrets
We provide meaningful explanations without exposing proprietary algorithms
We prioritize user rights over business secrecy
7. Security and AI Safety
7.1 AI System Security
Protecting AI Infrastructure:
Secure model storage and access controls
Encrypted data pipelines
Authentication and authorization for AI system access
Monitoring for unauthorized use or tampering
Adversarial Robustness:
Testing against adversarial attacks
Input validation and sanitization
Output verification and sanity checks
Anomaly detection in AI behavior
7.2 Preventing Misuse
Prohibited Uses of Our AI:
Discrimination or bias in hiring
Privacy violations or unauthorized surveillance
Manipulation or deception of candidates
Circumventing employment laws or regulations
Harassment or abuse of candidates
Enforcement:
Monitoring for misuse patterns
Client education on proper use
Suspension or termination for violations
Cooperation with authorities on illegal activities
7.3 Safety and Harm Prevention
Risk Assessment:
Regular evaluation of potential harms from AI use
Impact assessments for new features or models
Consideration of vulnerable populations
Mitigation strategies for identified risks
Incident Response:
Clear procedures for AI system failures or errors
Rapid response to bias or discrimination reports
Transparent communication about incidents
Corrective actions and lessons learned
8. Third-Party AI and Data Providers
8.1 Vendor Selection and Management
Due Diligence for Third-Party AI:
Evaluation of vendor AI ethics and fairness practices
Assessment of data sources and training methodologies
Review of privacy and security practices
Contractual requirements for compliance
Current Third-Party Providers:
Professional data enrichment
Public data search and discovery
Language models for analysis and summarization
Cloud infrastructure and hosting
Vendor Requirements:
Compliance with applicable data protection laws
Transparent data sourcing practices
Fairness and bias mitigation
Data Processing Agreements (DPAs) in place
Regular security and compliance audits
8.2 Control Over Third-Party AI
Our Responsibilities:
Select vendors with strong ethical practices
Ensure DPAs and security agreements are in place
Monitor vendor performance and compliance
Maintain ability to switch vendors if needed
Propagate data deletion requests to vendors
Limitations:
We rely on vendors' representations about their practices
We conduct reasonable due diligence but cannot audit vendors' internal systems
Third-party AI may have its own limitations and biases
We take responsibility for vendor selection and management
9. Compliance with AI Regulations
9.1 Current Regulatory Landscape
We actively monitor and comply with emerging AI regulations, including:
European Union AI Act:
Classification of our AI as "high-risk" for employment use
Enhanced transparency and documentation requirements
Bias testing and fairness validation
Human oversight mandates
U.S. State and Local AI Laws:
New York City Local Law 144 (automated employment decision tools)
Illinois Artificial Intelligence Video Interview Act
California AB 2029 (automated decision systems)
Other emerging state regulations
Sector-Specific Regulations:
EEOC guidance on AI in hiring
OFCCP compliance for federal contractors
Industry-specific employment laws
International Regulations:
UK AI regulation (following EU AI Act principles)
Canada AIDA (Artificial Intelligence and Data Act)
Other jurisdictions as applicable
9.2 Proactive Compliance Measures
What We Do Now:
Design AI systems with regulatory requirements in mind
Maintain documentation required by regulations
Conduct impact assessments for high-risk AI
Provide transparency and explainability
Implement human oversight and control
Future-Proofing:
Monitor regulatory developments worldwide
Adapt practices to meet new requirements
Engage with regulators and industry groups
Participate in AI ethics discussions
Regular compliance audits
9.3 Client Compliance Support
We Help Clients Comply With:
Employment laws and anti-discrimination regulations
Data protection laws (GDPR, CCPA, etc.)
AI-specific regulations in their jurisdictions
Industry-specific requirements
Resources We Provide:
Transparency documentation for AI systems
Data Processing Agreements (DPAs)
Privacy notices and candidate communication templates
Guidance on lawful basis and legitimate interest
Audit trails and record-keeping support
10. Accountability and Continuous Improvement
10.1 Governance Structure
Responsible Parties:
CEO: Ultimate accountability for ethical AI practices
CTO: Technical implementation of AI ethics principles
Data Protection Officer (DPO): Privacy and data protection oversight
Product Team: Embedding ethics in feature development
Compliance Team: Regulatory monitoring and adherence
Decision-Making:
Ethics considerations in all AI development decisions
Cross-functional review of new AI features
Regular executive review of AI practices and impacts
Board oversight of AI strategy and risk
10.2 Feedback and Reporting
Internal Feedback:
Team members encouraged to raise ethical concerns
Regular ethics discussions in development process
No retaliation for good-faith ethics concerns
External Feedback:
Candidate and client feedback channels
Third-party audits and assessments (as we scale)
Engagement with AI ethics researchers
Participation in industry working groups
Reporting Concerns:
Email info@shortlistd.io with "AI Ethics Concern"
Confidential reporting for sensitive issues
Investigation and response procedures
Transparent communication about resolutions (where appropriate)
10.3 Continuous Improvement
What We Track:
AI system performance and accuracy
Fairness metrics and bias indicators
Candidate satisfaction and feedback
Client satisfaction and hiring outcomes
Regulatory compliance status
Incident reports and resolutions
How We Improve:
Regular review of metrics and feedback
Iterative model improvements and updates
Updated policies based on lessons learned
Adoption of new best practices and technologies
Response to regulatory changes
Transparency About Progress:
Annual AI ethics report (planned)
Public disclosure of major changes to AI practices
Honest communication about limitations and failures
Ongoing dialogue with stakeholders
11. Contact and Questions
11.1 For Candidates
If you have questions about how AI was used in your recruitment process:
Email: info@shortlistd.io
Subject: "AI Processing Inquiry" or "Candidate Question"
Include: Your name and relevant job application details
Response Time: Within 30 days
11.2 For Clients
For guidance on ethical use of our AI tools:
Email: info@shortlistd.io
Subject: "AI Ethics Guidance"
11.3 For Researchers and Advocates
We welcome engagement with AI ethics researchers, advocates, and organizations:
Email: info@shortlistd.io
Subject: "AI Ethics Collaboration"
11.4 For Reporting Concerns
To report ethical concerns, bias, or potential violations:
Email: info@shortlistd.io
Subject: "AI Ethics Concern" or "Bias Report"
Confidential handling available upon request
12. Commitment and Disclaimer
12.1 Our Commitment
We are committed to:
Building AI that respects human dignity and rights
Preventing bias and discrimination in hiring
Transparent and explainable AI systems
Compliance with all applicable laws and regulations
Continuous improvement of our ethical practices
Accountability for AI outcomes and impacts
12.2 Honest Acknowledgment of Limitations
We Acknowledge That:
No AI system is perfect or completely free of bias
AI technology and best practices are rapidly evolving
We are learning and improving as we grow
Unexpected issues may arise despite our best efforts
We cannot guarantee perfect fairness or accuracy
Our Promise:
We will be transparent about limitations and failures
We will respond quickly to issues when they arise
We will continuously work to improve
We will prioritize ethics and fairness over commercial convenience
13. Policy Updates and Versioning
Review Schedule: This policy is reviewed and updated at least annually, or more frequently as:
AI regulations evolve
We deploy new AI capabilities
We learn from experience and feedback
Industry best practices change
Notification of Changes:
Material changes will be communicated via email and website notice
"Last Updated" date will be updated
Previous versions available upon request
Feedback Welcome: We actively seek input on this policy. Email info@shortlistd.io with "AI Policy Feedback" to share your thoughts.
This Ethical AI Policy is integral to our Terms of Service and Privacy Policy and should be read in conjunction with those documents.