AI Acceptable Use Policy

1. Purpose and Scope

This AI Acceptable Use Policy governs the use of artificial intelligence and machine learning technologies within Virvell's employment screening platform. It applies to all AI-powered features including:

• Voice AI for conducting reference check conversations
• Pre-screening interview assessments
• Transcript analysis and insights generation
• Report generation and summarization

2. Core Principles

2.1 Human Decision-Making Authority

AI systems at Virvell are designed to inform and assist human decision-makers, never to replace them. Specifically:

• All employment decisions are made by qualified humans. Our AI provides analysis and insights; humans make hiring decisions.
• AI-generated assessments are advisory only and should be considered alongside other factors.
• Users are responsible for making their own independent judgments about candidates.

2.2 What Our AI Does NOT Do

2.3 What Our AI Does

Our AI systems are designed to:

• Analyze response quality: Evaluate the clarity, completeness, and relevance of interview responses
• Summarize information: Provide structured summaries of reference conversations
• Identify patterns: Highlight themes and notable observations from transcripts
• Generate reports: Create comprehensive documentation for human review
• Support workflow: Help organize and prioritize information for human decision-makers

3. Assessment Language and Labels

3.1 Response Quality vs. Candidate Evaluation

All metrics and scores in our system evaluate response quality, not candidate worth or employability:

• Communication scores measure how clearly responses were articulated
• Completeness scores measure how thoroughly questions were answered
• Relevance scores measure how well responses addressed the questions asked

3.2 Workflow Status Labels

Our system uses workflow status labels that indicate next steps for human review, NOT employment recommendations:

• "Proceed" - Responses support moving to the next workflow stage
• "Review" - Responses warrant additional human discussion
• "Hold" - Responses require further clarification before proceeding

These labels help organize your review queue. They do not indicate whether a candidate should be hired.

4. Pre-Screening and Eligibility Checks

4.1 Eligibility Criteria

Pre-screening interviews collect information about basic job requirements such as:

• Work authorization status
• Availability and schedule compatibility
• Required certifications or licenses
• Location and travel requirements

4.2 Human Review Requirement

By default, all pre-screening results require human review before any candidate status changes. Our system:

• Collects and organizes eligibility information
• Flags responses that may not meet stated job requirements
• Presents all information to human reviewers for decision-making

Employers may optionally configure automated workflow notifications, such as interview booking emails for candidates whose responses support advancement, or professional not-proceeding emails for candidates whose responses require further clarification. These notifications execute only after an employer-configured delay period (default 24 hours), during which human review, override, and cancellation remain available through the dashboard. These features are opt-in, off by default, and do not replace human decision-making authority. The AI assigns workflow labels; the employer decides whether and how to act on them.

5. Reference Check Analysis

5.1 Reference Feedback Summaries

Our AI analyzes reference conversations to provide:

• Thematic summaries of reference feedback
• Identification of consistent patterns across references
• Structured presentation of qualitative information

5.2 Signal Quality and Consistency Analysis

Our system may identify patterns that warrant attention, such as:

• Inconsistencies between different references
• Unusual response patterns
• Areas where additional clarification may be helpful

These insights are provided for human consideration, not as definitive judgments.

5.3 What "cross-verification" means

When Virvell describes "cross-verification" or surfaces verified facts, contradictions, and gaps in customer-facing reports, this refers to module-level comparison across our information sources: resume parsing, AI pre-screen interview, AI voice reference checks, third-party background verification, and the job requirements provided by the customer. The platform identifies where these sources agree, where they conflict, and where information is missing — and presents these patterns to human reviewers for evaluation.

In addition to the module-level comparison above, Virvell offers statement-level source attribution on accounts where it has been enabled. Where active, the platform links individual factual statements it extracts from AI call transcripts and from submitted resume text back to their origin: the specific transcript utterance (including its position in the call) or the specific location in the resume text. Each link is checked by the platform before it is stored — the quoted statement must actually appear at the recorded location in the stored source. Statements that cannot be confirmed against their source are flagged for human review and are never presented as attributed. Statement-level attribution currently covers call transcripts and resume text; facts sourced from third-party background verification are cited at the level of the verification report, not to a location within it.

On accounts where statement-level attribution has not been enabled, "source attribution" continues to refer to the module-level comparison described above. Statement-level attribution is display of evidence and its origin — it does not add any assessment of the candidate, and all interpretation remains with human reviewers.

As source-attribution capabilities change, this section is updated to reflect what the platform actually does at that time.

6. Data Protection and Privacy

6.1 AI Data Handling

Our AI systems process data in accordance with our Privacy Policy and applicable data protection laws. Key principles include:

• Minimizing data retention to what is necessary
• Securing all personally identifiable information
• Providing transparency about how AI processes information

6.2 Consent and Disclosure

All parties involved in AI-powered interactions are informed that:

• They are interacting with AI-powered voice technology
• Conversations may be recorded and analyzed
• They may decline to participate

7. User Responsibilities

Users of Virvell's platform agree to:

• Make independent decisions: Use AI-generated insights as one input among many, not as the sole basis for employment decisions
• Comply with applicable laws: Ensure their use of AI-assisted screening complies with all applicable employment, privacy, and anti-discrimination laws
• Apply human judgment: Review AI-generated assessments critically and apply professional judgment
• Maintain fairness: Not use AI outputs in ways that discriminate against protected classes

8. Continuous Improvement

We are committed to continuously improving our AI systems to:

• Reduce potential biases in analysis
• Improve accuracy and helpfulness of insights
• Maintain alignment with evolving best practices and regulations
• Respond to user feedback and concerns

9. Questions and Concerns

10. Policy Updates

This policy may be updated periodically to reflect changes in our AI capabilities, applicable laws, or industry best practices. Material changes will be communicated to users through our platform or via email.

This policy reflects our commitment to responsible AI use in employment screening. We believe that AI should enhance human capabilities while preserving human authority over important decisions that affect people's lives and livelihoods.