AI Transparency Notice & Disclosure

1. How We Use AI

KybaBox uses automated processing (specifically Large Language Models, Computer Vision, and Voice-to-Text) to assist you in organizing your belongings. When "AI Mode" is enabled, the system performs the following tasks:

• Object Recognition: Identifying items within your uploaded photos.
• Categorization: Suggesting box names, destination room, and inventory tags (e.g. fragile or heavy).

2. You're in Control

KybaBox AI is designed as a recommendation engine, not a final decision-maker.

• Review & Edit: All AI-generated suggestions are presented to you for review. No data is finalized until you click "Save."
• Override: You retain complete control to edit, reject, or delete any AI-generated content.
• Opt-Out: You can disable AI features at any time via your account settings.

3. Data Privacy & Regional Processing

KybaBox is built on secure, enterprise-grade AI infrastructure and is available exclusively in the United States, European Union (EU), and United Kingdom (UK). All AI requests are dynamically routed through Google Cloud's Gemini Enterprise multi-region infrastructure based on your location:

• United States Users: Routed to and processed within the US multi-region infrastructure.
• EU and UK Users: Routed to and processed strictly within the EU multi-region infrastructure.

We enforce strict architectural controls to maximize your privacy across all regions:

• No-Cache Policy: Infrastructure-level prompt caching is explicitly disabled. Every AI request is processed as a unique, independent event.
• No Model Training: Your media (photos and voice recordings) and text inputs are never used by KybaBox or third-party providers to train or fine-tune AI models.
• Ephemeral Safety Buffer: To enforce platform safety and abuse prevention, Google Cloud retains inputs in a secure, temporary buffer for up to 30 days. This data is subject to automated deletion, is restricted by "no-eyes-on" access controls, and is never used for training or human evaluation under normal operations.

4. Safety and Quality Control

Our infrastructure includes built-in safety filters provided by our technology partners to block the processing of harmful or inappropriate content.

• Prohibited Use: KybaBox is a utility for home organization. It is not designed or intended for "high-risk" decisions, such as determining insurance eligibility, legal valuations, or creditworthiness.
• Active Monitoring: We actively monitor the quality of our AI suggestions to ensure the system remains a helpful and neutral tool for all users. We analyze aggregated, anonymized accept/reject rates across different categories to identify where the model may be underperforming, ensuring a neutral and reliable experience without ever accessing user media.

5. Accuracy and Feedback

While our AI is highly advanced, it may occasionally misidentify items based on lighting, camera angle, image clarity, or other factors.

• Human-in-the-Loop: To ensure your inventory is perfect, KybaBox puts you in the driver's seat. You'll always have the opportunity to review and refine AI suggestions, ensuring your digital box details are accurate.
• Technical Support: If you experience a persistent technical failure, you may report it via our Help contact form.

Our Commitment: We believe AI should be a helpful assistant, not a data harvester. If you have questions about our privacy configurations, we're happy to chat.

Data Controller: KybaBox, LLC 5900 Balcones Dr #29996 Austin, TX 78731

Contact: Help contact form