Moderation Framework Overview

We employ a multi-layered moderation approach combining automated safeguards, application-level restrictions, continuous monitoring, and human review to proactively prevent, detect, and remediate prohibited content.

1. AI Model Safety Controls

Our platform relies on third-party and proprietary AI models that include built-in safety and compliance mechanisms. These models are designed to:

• Block the generation of illegal, explicit, abusive, or otherwise harmful content
• Enforce internal content safety standards aligned with applicable laws and industry best practices
• Prevent outputs that violate widely accepted content policies, including those related to violence, sexual exploitation, fraud, hate speech, or other prohibited categories
  

These safeguards operate at the model level prior to content generation.

2. Application-Level Input Restrictions

In addition to model-level protections, we implement application-side controls that restrict content before it is submitted to AI models. Specifically:

• User prompts and requests are programmatically analyzed for restricted keywords, phrases, and patterns associated with illegal or harmful intent
• Requests identified as attempting to generate prohibited content are blocked and not forwarded to the AI model
• Repeated attempts to bypass these restrictions are flagged for further review
  

This layer serves as a preventative control to reduce misuse of the platform.


3. Automated Post-Generation Content Scanning

All generated content is subject to ongoing automated monitoring after creation. We use AI-based scanning systems to:

• Review generated outputs for indicators of illegal, harmful, or policy-violating content
• Detect content that may have evaded earlier preventative checks
• Flag high-risk content for immediate action
  

If content is determined to violate policy, it is promptly removed, and associated user accounts may be restricted or suspended.

4. Human Moderation and Manual Review

We maintain a dedicated content moderation team responsible for manual review and enforcement. This team:

• Conducts routine audits of generated content
• Reviews content flagged by automated systems
• Evaluates user behavior patterns for abuse or circumvention attempts
  

Confirmed violations result in content removal and may lead to account suspension or termination, depending on severity and recurrence.

5. Enforcement and Remediation

When prohibited content is identified, we take appropriate corrective actions, which may include:

• Immediate deletion of the content
• Temporary or permanent suspension of user accounts
• Escalation for further internal review where required
  

We maintain internal logs of moderation actions to support compliance reviews and audits.

6. Ongoing Review and Improvement

Our content moderation processes are continuously evaluated and updated to:

• Align with evolving legal requirements
• Reflect changes in Card Brand standards and bank policies
• Improve detection accuracy and response times