Almost Leaked User Data with AI Feature – A Wake‑Up Call from API Terms of Service

Description and use

The developer realized that free Gemini API may use user data for training and added warnings, UI notices, and code comments to prevent accidental leakage of confidential information.

Problem

User‑generated text entered into a free‑tier AI API can be used by the provider for product improvement, risking accidental leakage of confidential data.

Outcome

Users are now aware that their input may be used for training; the developer mitigated accidental data leakage by adding warnings and documentation.

Workflow

• 1. Added prominent disclaimer in manual and FAQ.
• 2. Updated UI to display warning link near input form.
• 3. Documented risk in function docstring and added runtime comment.
• 4. Ensured API key is sourced from environment variables for security.

Evidence

Content submitted through the free tier APIs (including prompts and responses) may be used to improve Google's products and services, such as generative AI models.

Source