At Sabian, we build artificial intelligence systems designed to serve critical industries, national programs, and strategic operators who cannot rely on guesswork, opacity, or uncontrolled automation. Our AI technology operates with a strict commitment to safety, sovereignty, transparency, and respect for human decision-making. This Responsible AI Statement outlines the principles guiding how Sabian designs, deploys, and governs its systems. 1. Human-Centered by Design Sabian AI Technology supports and augments engineers, operators, analysts, and decision-makers — never replaces them. • All recommendations require human validation. • Sabian does not autonomously control machinery, adjust valves, change reagent dosing, or execute process actions. • Operators remain in full command of every operational decision. The system is a decision-intelligence layer, not an automated controller. 2. Safety in Industrial and National Environments Our technology is deployed in high-stakes operational environments: rare-earth processing plants, mineral supply chains, national intelligence frameworks, and strategic procurement ecosystems. To ensure safety: • Sabian models plant behavior without intrusive control. • It detects instability early to protect personnel, equipment, and production. • It maintains strict boundaries to avoid interfering with critical operations. • All outputs are designed to reduce operational risk, not increase it. 3. Privacy, Confidentiality & Sovereign Data Protection Sabian treats all plant data, operational logs, personnel information, and strategic intelligence as protected assets. We commit to: • Zero data resale. • Zero data sharing without contractual authorization. • Zero third-party access beyond approved security partners. • Strong encryption at rest and in transit. • Data residency compliance with sovereign requirements. • Secure deletion protocols upon request or contract termination. Sabian is engineered so your data remains yours. 4. No Use of Customer Data for Model Training Sabian does not use client data to train global models or commercial AI products. Each deployment has its own protected model context — isolated, encrypted, and customized to your plant or organization. Your proprietary chemistry, flow sheets, equipment signatures, and operational behavior are never generalized or shared. 5. Transparency of Recommendations Every AI-generated insight includes context explaining: • the conditions that triggered the alert • the patterns or behaviors detected • the variables contributing to the prediction • the level of confidence • the historical backing for the insight Users always understand why Sabian is making a recommendation never a “black box.” 6. Bias Mitigation and Data Integrity Sabian’s intelligence architecture is designed to minimize bias in industrial and geopolitical analysis. • Data sources are validated for reliability and consistency. • Conflicting data is flagged, not smoothed or altered. • Model predictions are cross-checked with physical, chemical, and mechanical constraints. • Rare-earth market intelligence is adjusted for known geopolitical distortions. Sabian reports what is true, not what is convenient. 7. Restricted Use Policy Sabian prohibits the use of its systems for: • autonomous weapons development • harmful surveillance • activities violating international law • discriminatory decision-making • environmental or safety negligence • operations that endanger personnel or communities Sabian is built for strategic advantage, not misuse. 8. Continuous Review and Improvement Our Responsible AI practices evolve with scientific research, industry standards, regulatory frameworks, and client needs. We regularly review: • safety safeguards • data governance • model performance • transparency methods • risk assessments Sabian is committed to industry-leading AI governance.