ACHNET AI Ethical Principles Our Framework for Building Responsible AI in Hiring

Introduction At ACHNET, we donʼt just build AI — we build responsible, auditable, and human-centered systems that help organizations hire more effectively. iJupiter©, our proprietary AI Agent, is designed to accelerate hiring timelines while upholding the highest standards of ethics, fairness, transparency, and compliance. As AI technology advances, we recognize our responsibility to ensure it supports diverse workforces, respects privacy, and enhances human decision-making — not replaces it. This document outlines the ethical guardrails embedded into iJupiter©ʼs lifecycle, from design and deployment to governance and oversight.

Our Five Guiding Ethical Commitments

1. AI Must Enhance Human Judgment, Not Replace It iJupiter© is built as a decision-support engine, not a decision-maker. Our tools surface insights that enable hiring managers to make smarter decisions — faster. We empower humans with structured data, allowing them to combine AI-backed recommendations with interpersonal evaluation and team-fit judgment.

Final hiring decisions are always made by people, with iJupiter© providing ranked candidates, competency insights, and performance indicators.

2. Fairness and Inclusion Are Engineered into the Model Fairness isnʼt a feature — itʼs a foundation. ACHNET embeds bias mitigation into every stage of iJupiter©ʼs design and validation:

Use of Behaviorally Anchored Rating Scales (BARS) for training supervision
Exclusion of irrelevant features such as facial expressions, tone of voice, or background context from evaluations
Regular adverse impact testing against protected categories (e.g., gender, ethnicity, age, disability)

We proactively audit and recalibrate models to ensure they operate equitably across diverse candidate populations.

3. Privacy by Design and Default ACHNET adheres to global data protection laws including GDPR, UK GDPR, and CCPA. Privacy is engineered into our workflows from the ground up:

Minimal data collection: We collect only what is necessary for job-related evaluation.
Encryption at rest and in transit: All candidate data is secured throughout its lifecycle.
Controlled access: Only authorized personnel may access candidate records.
Candidate rights: Candidates may request deletion of their personal data via support@achnet.com.

Our clients act as data controllers, and we operate as a data processor, strictly following contractual obligations.

4. Transparency and Explainability at Every Touchpoint We believe in AI you can understand and trust. Thatʼs why we provide:

Structured scoring reports for recruiters with competency-level breakdowns
Documentation of evaluation logic, highlighting the attributes analyzed and the models applied
Optional Candidate Feedback Reports shared at the discretion of our clients

We also publish public-facing documents such as our AI Explainability Statement and Model Risk Guidelines, so businesses know exactly how iJupiter© works — and why

5. Continuous Validation and Governance ACHNET does not launch and forget. iJupiter© is:

Validated using performance outcomes and predictive scoring
Tested for bias and revalidated annually
Monitored live for score drifts or anomalies

All AI deployments are reviewed by our AI Ethics Governance Council, which includes members from product, data science, compliance, and external legal advisors. This council oversees policy changes, fairness audits, and model risk scoring using our proprietary AI Trustworthiness Framework™

How ACHNET Designs for Inclusion Rather than using off-the-shelf algorithms, iJupiter© assessments are built using a repeatable, defensible methodology:

Define the performance criteria for each job role.
Build assessment prompts that elicit role-relevant behavioral responses.
Score training data using BARS methodology supervised by domain experts.
Train machine learning models to predict role-specific success, not generic traits.
Remove features or model paths that contribute to adverse impact without improving validity.
Test thoroughly before launch, and monitor continuously post-deployment.

Promoting Diversity Through Design ACHNETʼs commitment to diversity isnʼt just about candidate treatment — itʼs reflected in our:

Team structure Cross-functional teams include diverse backgrounds in data science, I/O psychology, law, and DEI.

Data diversity Our models are trained on anonymized, multi-industry, multi-region datasets.

Customer controls Clients can customize models and thresholds to align with local employment laws and cultural norms.

Closing Note: Responsible AI is everyoneʼs job ACHNET invites collaboration — not just from clients, but from regulators, researchers, and civil society — to shape the future of responsible hiring. We are committed to improving our practices, increasing transparency, and sharing learnings that advance the field.