AI and Automated Decision Systems Policy

1. Purpose

This policy describes how Twenty One Nexus designs and provides AI-assisted recruiting features and the responsibilities of customers using them. It is intended to support transparency and responsible human decision-making, not to represent universal legal compliance for every use case.

2. Covered features

Resume and profile parsing, extraction, search and summarization.

Candidate-to-job matching, scoring, ranking and recommendations.

Candidate rediscovery and talent-pool creation.

Sourcing suggestions, enrichment and outreach drafting.

Job-description, interview-question and message generation.

Workflow automation and AI-assisted responses.

3. Human oversight

AI outputs are recommendations or drafts. Customers must apply meaningful human review, consider relevant context outside the system, allow correction of inaccurate data, and avoid automatic rejection or other consequential actions unless independently validated, legally permitted, transparently disclosed, and appropriately governed.

4. Prohibited inputs and factors

Customers must not intentionally use race, color, religion, sex, pregnancy, sexual orientation, gender identity, national origin, age, disability, genetic information, veteran status, or other protected characteristics as selection criteria, except where law expressly requires or permits processing for a legitimate compliance purpose with appropriate safeguards. Customers should not use proxies designed to recreate protected traits.

5. Explainability and records

Display major factors contributing to a recommendation where technically feasible.

Identify missing or uncertain information and distinguish extracted facts from generated inferences.

Maintain logs of model/version, date, inputs, output, user changes, overrides and final human action where configured.

Provide exportable information reasonably needed for customer audits and investigations.

6. Testing and monitoring

Provider should maintain documented testing appropriate to the feature, including accuracy, robustness, security, and risk of material performance differences. Material changes should be versioned and assessed. Customers remain responsible for validating job-related criteria and monitoring outcomes in their own applicant population.

7. Regulated employment tools

A customer's configuration or use may fall within laws governing automated employment decision tools or automated decision-making technology. Customers are responsible for determining applicability, completing required notices and assessments, offering accommodations or alternative processes, and obtaining any required independent bias audit. Provider will supply reasonably available technical information under the DPA or a compliance addendum.

8. Candidate rights and contact

Customers should provide candidates a clear channel to request information, correct inaccurate data, seek accommodation or alternative assessment, and request human review where required. Questions about Provider practices may be sent to [email protected].