Mahjee AI Governance Policy

Effective Date: October 24, 2024

1. Overview of AI Governance

At Mahjee, we recognize that Artificial Intelligence (AI) technology brings unprecedented opportunities to transform industries, optimize decision-making, and enhance customer experiences. However, it also introduces unique risks and challenges that require a comprehensive governance framework. AI Governance at Mahjee encompasses a structured approach to the development, deployment, monitoring, and continuous improvement of AI systems to ensure they meet intended business objectives, are used responsibly, and comply with evolving legal, ethical, and regulatory standards.

Our AI Governance framework integrates AI risk management into existing data privacy, cybersecurity, and compliance programs, thereby avoiding the creation of a separate governance regime. This approach ensures a consistent set of policies and practices across the organization. We align our governance strategy with the NIST AI Risk Management Framework (RMF), which focuses on the principles of “Map, Measure, Manage, and Govern.” These principles guide us in identifying, assessing, mitigating, and continuously monitoring AI-related risks across the AI lifecycle.

2. AI Governance Board

The AI Governance Board at Mahjee is a cross-functional committee that includes representatives from various departments such as product development, compliance, marketing, sales, legal, senior management, and user advocacy. This diverse representation ensures that different perspectives are considered when making decisions related to AI governance. The responsibilities of the AI Governance Board include:

• Identifying and Evaluating AI Risks: The Board is responsible for identifying potential risks associated with the development and deployment of AI systems. This includes technical risks (e.g., algorithmic bias, data quality), operational risks (e.g., system failures, unintended consequences), and legal risks (e.g., compliance with data protection laws).
• Establishing Standards, Metrics, and Risk Controls: The Board sets the standards for AI system development, including performance metrics, data quality requirements, and risk control measures to ensure that AI systems meet Mahjee’s ethical and compliance standards.
• Regular Assessments and Compliance Monitoring: The Board conducts periodic assessments of AI systems to ensure that they comply with regulatory requirements, organizational policies, and industry best practices. This includes monitoring for changes in AI regulations and adapting our governance approach accordingly.
• Implementing Technical, Legal, and Operational Protections: The Board ensures that Mahjee’s AI systems are protected against risks by implementing a range of safeguards, including encryption, access controls, data anonymization, and robust incident response protocols.
• Adapting to Evolving Legal Requirements: The Board stays abreast of evolving AI regulations, such as the EU AI Act, state-level AI laws in the United States, and international AI governance frameworks. It ensures that Mahjee’s AI systems are continuously updated to comply with these requirements.

3. Trustworthy AI Standards

To ensure that our AI systems operate ethically and effectively, Mahjee adheres to a set of Trustworthy AI standards that cover various aspects of AI development and deployment. These standards are based on established principles of AI ethics and risk management, ensuring that our AI systems are robust, reliable, and aligned with societal values. The key Trustworthy AI standards include:

• Validity: AI systems must be designed to meet the requirements for their intended use cases. This includes validating that the algorithms used are suitable for the problem being solved and that the training data is representative, high-quality, and obtained from authorized sources. Mahjee performs rigorous testing and validation to ensure the system’s accuracy and appropriateness.
• Reliability: AI systems must perform consistently under different conditions, including changes in data inputs, updates to the model, and variations in operational environments. Reliability testing is conducted regularly to ensure that the system maintains its accuracy and performance over time.
• Safety: Our AI systems are designed to avoid risks that may harm life, health, property, or the environment. Safety assessments include evaluating the potential impact of AI decisions on users and communities, particularly in high-stakes domains such as healthcare or autonomous systems.
• Security: AI systems are secured against threats such as data poisoning, adversarial attacks, and unauthorized access. Security measures include encryption, access controls, anomaly detection, and regular security audits to identify and address vulnerabilities.
• Resilience: AI systems are designed to recover and return to normal functionality after adverse events such as cyber-attacks, system failures, or data corruption. This includes implementing redundancy, failover mechanisms, and automated recovery protocols.
• Accountability: Mahjee takes responsibility for the outcomes of AI system decisions. This includes establishing clear ownership and oversight for AI models, documenting decision-making processes, and implementing mechanisms for addressing errors or grievances.
• Transparency: We are committed to providing transparency regarding our AI systems’ design, operation, limitations, and potential risks. We disclose information to stakeholders about the data used, model performance, and any inherent limitations or uncertainties in the system.
• Explainability: Mahjee ensures that the mechanisms and outputs of AI systems can be clearly described and understood by stakeholders, including end-users and regulatory authorities. We prioritize creating models that offer interpretable insights rather than relying solely on “black-box” algorithms.
• Privacy Enhancement: Our AI systems are built with privacy in mind. This includes incorporating data protection measures such as differential privacy, data anonymization, and minimizing data retention periods. We strive to balance the benefits of AI data processing with the need to respect individual privacy.
• Human-Centric Design: Mahjee’s AI systems are developed with a focus on human values, such as fairness, justice, and respect for fundamental freedoms. This includes avoiding discrimination, enhancing accessibility, and promoting user autonomy.
• Bias Management: We actively work to identify, mitigate, and manage bias in AI systems. This involves diverse data sourcing, fairness testing, and implementing safeguards to reduce bias propagation through model updates.

4. Compliance through Use Case Analysis

Before deploying any AI system, Mahjee conducts a thorough analysis of the use case to understand the potential benefits, risks, and regulatory requirements. The use case analysis considers:

• Business Purpose and Impact Assessment: Understanding the specific business goals for deploying the AI system and evaluating the potential impact on customers, employees, and other stakeholders. This includes assessing whether the AI will enhance existing processes, create new opportunities, or present potential risks.
• Position in the AI Supply Chain: Identifying Mahjee’s role in the AI ecosystem, whether as a developer, integrator, or end-user of third-party AI solutions. This helps determine the level of responsibility and accountability required.
• Risk Assessment of High-Risk or Consequential AI Applications: Evaluating if the AI system will be used in critical domains such as healthcare, financial services, or public safety, where decisions may have significant consequences for individuals or society.
• Data Rights and Compliance: Ensuring that the data used for AI training and inference complies with legal and contractual obligations, including data protection laws, intellectual property rights, and data-sharing agreements.
• Vendor Risk Assessments: For AI systems sourced from third-party vendors, Mahjee conducts risk assessments to ensure that the vendor’s AI practices align with our standards. This includes evaluating the vendor’s approach to data governance, algorithmic fairness, and compliance with regulatory requirements.

5. Roles and Responsibilities

Mahjee’s AI governance framework assigns clear roles and responsibilities for overseeing AI systems. Key roles include:

• AI Governance Board: The Board is responsible for establishing governance standards, reviewing compliance reports, and providing oversight for AI-related initiatives. They ensure that governance practices evolve with technological advancements and regulatory changes.
• Data Protection Officer (DPO): The DPO ensures that AI practices comply with data protection regulations and advises the organization on privacy risks associated with AI. The DPO works closely with the AI Governance Board to monitor data-related risks.
• AI Ethics Officer: The AI Ethics Officer is responsible for ensuring that Mahjee’s AI practices align with the organization’s ethical principles. They provide guidance on ethical dilemmas and oversee fairness, bias, and discrimination issues.
• Legal and Compliance Team: The Legal and Compliance team provides advice on regulatory requirements, intellectual property matters, and contractual obligations related to AI. They help ensure compliance with local and international laws.
• Data Scientists and AI Engineers: Data scientists and AI engineers are responsible for developing and validating AI models according to the governance standards. They ensure models are trained on high-quality data and meet the organization’s performance and fairness requirements.
• Risk Management Team: The Risk Management team identifies, assesses, and mitigates risks associated with AI. They work closely with other teams to implement risk controls and monitor the ongoing performance of AI systems.

6. Continuous Improvement and Adaptation

Mahjee is committed to continuously improving its AI Governance framework to keep pace with emerging trends, regulatory developments, and evolving societal expectations. The AI Governance Board regularly reviews the policy and procedures to ensure they remain relevant and effective. This commitment to continuous improvement includes:

• Ongoing Training and Education: Mahjee provides regular training to employees on AI governance principles, ethical considerations, and compliance requirements to ensure everyone understands their role in responsible AI development.
• Regular Policy Reviews: The AI Governance Policy is reviewed at least annually or whenever significant changes occur in AI technology or regulatory standards. This ensures that the policy remains up to date and aligns with best practices.
• Stakeholder Engagement: Mahjee engages with external stakeholders, including customers, regulatory bodies, and industry experts, to gather feedback and stay informed of emerging AI trends and regulatory developments.

7. Reporting and Accountability

Mahjee’s AI governance framework emphasizes transparency and accountability. The AI Governance Board is responsible for overseeing the implementation of the governance policy and reporting on its effectiveness. This includes documenting compliance efforts, conducting internal audits, and publishing relevant findings to stakeholders.