How do I choose an AI model with good privacy policies?
What to Look For in a Privacy Policy
Start by reading the privacy policy and terms of service. A good policy will explicitly say what data the AI collects (e.g., prompts, usage logs), why it's collected, how long it's kept, and with whom it's shared. Avoid vague language like 'we may share data with partners.'
Check if the provider offers data controls: Can you delete your history? Can you opt out of training? Does the policy apply to all features or just some? For example, some AI models let you disable chat history, which prevents your conversations from being used for training.
- Look for a dedicated privacy policy, not just a general terms page.
- Check for data retention periods (e.g., 30 days, 90 days).
- See if you can opt out of data collection for model training.
- Look for compliance with regulations like GDPR or CCPA.
- Check if data is encrypted in transit and at rest.
Comparing Providers
Different AI providers have different defaults. Some, like OpenAI, allow you to opt out of training via settings, while others may use your data by default. Enterprise or API plans often have stricter data handling than free consumer versions.
Open-source models you run locally (e.g., Llama) keep data on your device, but you're responsible for security. Cloud-based open-source models may still log data, so check the host's policy.
Common mistakes
- Assuming that because a model is popular, its privacy policy is strong; popularity doesn't guarantee good data practices.
- Thinking that deleting your chat history automatically removes all data; some providers retain anonymized logs.
- Overlooking that free tiers often have weaker privacy protections than paid or enterprise plans.
