AI Safety at KazeAI

We are committed to building AI that is safe, trustworthy, and beneficial for every Rwandan.

Content Moderation

All AI responses are filtered through our content moderation system that blocks harmful, illegal, or abusive content. Banned words and topics are configurable by platform administrators.

Medical & Legal Disclaimers

KazeHealth and KazeLaw modules clearly display disclaimers that AI responses are not professional medical or legal advice. Users are directed to qualified professionals for critical decisions.

Transparency

We are transparent about the fact that users are interacting with AI, not human professionals. All AI-generated content is clearly marked as such.

Data Privacy

Conversations are private and encrypted. We do not share personal conversation data with third parties. Users can delete their data at any time.

Human Oversight

Our moderation team reviews flagged content. Users can flag any response they believe is harmful, inaccurate, or inappropriate.

Bias Mitigation

We actively work to identify and reduce bias in our AI systems, particularly in culturally sensitive topics relevant to Rwanda.

How to Report Harmful Content

If you encounter content that you believe is harmful, inaccurate, offensive, or inappropriate:

  1. Click the flag icon on any AI message to report it directly
  2. Email our safety team at safety@kazeai.rw
  3. Use the Contact page to describe the issue in detail

Limitations of AI

Our AI systems have limitations you should be aware of:

  • AI can make factual errors — always verify important information
  • AI responses may not reflect the most current laws, prices, or medical guidelines
  • AI does not replace professional doctors, lawyers, or financial advisors
  • Kinyarwanda AI quality is continuously improving but may have gaps
  • AI should not be used for emergency situations — call 112 for emergencies

Emergency Services

KazeAI is not an emergency service. For medical emergencies, call 912. For police, call 113. For fire, call 111. Do not rely on KazeAI for life-threatening situations.