AI Privacy Explained: What Happens to What You Type

The question that matters: training vs. storage vs. retention

These are three separate things, and providers handle each differently:

• Training use — whether your conversations may be used to improve future versions of the model.
• Storage — whether your conversations are stored at all, and for how long.
• Retention/deletion — whether and how you can delete your data, and how long deletion actually takes to take effect.

Check the settings, not just assumptions

Most major consumer AI chat products (ChatGPT, Claude, Gemini) offer a setting to opt out of having your conversations used for model training, though the exact mechanism and defaults differ and change over time — check the current settings in the tool you're using rather than relying on general assumptions. Business/enterprise tiers of these same products typically come with stronger data-handling guarantees (including not training on your data by default) as part of the paid offering.

A simple rule for sensitive information

Treat any AI chat input as you would an email to a third-party vendor: don't paste information you wouldn't be comfortable a vendor's staff or systems processing, unless you've specifically confirmed the tool's business-tier data handling terms cover your situation. This applies to things like unreleased business plans, client personal data, credentials, and health or financial information.

Where to actually check

Each tool's own privacy policy and (for business use) Data Processing Agreement are the authoritative source — not general articles like this one, since terms change. Every tool page on this site links to the provider's own site so you can review their current terms directly.