7 Principles for Real Privacy in the Age of AI

"Protect your privacy" is easy to say, but real protection has to be concrete. Here are seven principles for keeping your data yours while using AI tools.

1. Before uploading, ask: does this really need to go to a server?

Many tasks — conversion, summarization, transcription — can be done on your own device. Not uploading files that don't need to be uploaded is the best defense there is.

2. Provide only the minimum

Strip out information the task doesn't need — names, contact details — before you feed it in. The AI only processes what you give it.

3. Check the retention policy

If uploading is unavoidable, find out how long the service keeps your data and whether it's used for training.

4. Prefer on-device options

Given the same functionality, choose the tool that processes in your browser or on your device first.

5. Separate file handling from accounts and tracking

"We don't upload your files" is not the same as "we don't track anything." Ad and analytics cookies are a separate matter — check and manage them on their own.

6. Manage share links and outputs too

The moment you post a result to a share link, that data is back out in the world. Keep track of where your outputs get saved and shared as well.

7. Insist on verifiability

A claim of "we don't send your data" is only trustworthy if you can verify it. Check the network requests in your browser's developer tools, or look for open standards (WebGPU, WASM) under the hood.