AI Usage Policy

• Productivity: drafting, summarizing, translation (English, Bangla, Banglish), coding assistance.
• Enterprise analytics with human-reviewed outputs.
• Customer support augmentation with clear AI disclosure.
• Research and internal knowledge search over your own data.

• Fully automated consequential decisions about people (employment, credit, housing, insurance, healthcare, education, immigration, benefits, or criminal justice) without meaningful human review.
• Biometric categorization, emotion inference in the workplace or schools, or social scoring — the practices prohibited under Article 5 of the EU AI Act.
• Generating non-consensual intimate imagery, CSAM, or content that sexualizes minors.
• Targeted disinformation, election manipulation, or impersonation of real people without consent.
• Development of weapons, malware, or systems designed to cause physical harm.
• Circumventing safety filters or extracting model weights.

• Users interacting with an AI agent are informed they are speaking with AI.
• AI-generated content is watermarked or labeled where feasible and required by law.
• Every AI output carries a Veracity score and Analytics-Ladder tag (Descriptive / Diagnostic / Predictive / Prescriptive) as per our internal doctrine.

High-impact workflows must include a human-in-the-loop review step. Workspace owners are responsible for configuring approvals, escalation, and override paths appropriate to their industry.

• Customer workspace content is not used to train foundation models.
• Prompts and outputs may be processed by third-party model providers routed via the OMS AI Nexus AI Gateway under contractual confidentiality.
• Do not submit regulated data (health, financial, biometric, or children's data) unless your subscription includes the corresponding controls.

Model outputs are monitored for safety violations. Report harmful outputs, jailbreak attempts, or safety incidents to omsit.official@gmail.com. We will respond within 5 business days and, where appropriate, notify affected users and regulators.

Violation of this policy may lead to output blocking, feature suspension, or account termination. Repeated or severe violations may be reported to authorities where required by law.