AI needs the information, not the person's identity
A doctor wants to summarize a patient report, a specialist wants to organize session notes, and an employee wants to analyze a client spreadsheet. All useful work, but the file contains names, numbers, and private details from people who trusted you. The rule is simple: AI needs the information to help you, not the identity of the person it belongs to.
Remove identifying details before uploading
- Name: Replace it with a code, such as “Client 1” or “Case A.”
- Phone number, address, and national ID number: Delete them, because it does not need them.
- Dates: If the month is enough, remove the day.
- Unusual details: A rare job or small location may identify someone even without a name.
In a spreadsheet, replace the names column with numbers before uploading, and keep the number-to-name mapping table only with you.
Adjust the settings
ChatGPT, Claude, and Gemini settings include an option controlling whether your conversations are used to improve models, as well as temporary chats that are not saved in history. Review them in your account.
If you work in a company or hospital, ask about the account the organization provides and its AI usage rules before uploading any work file.
Consent first
If you will record a session, medical appointment, or call to transcribe it, get the person's consent first. If you will share a real situation in content, change the details and get their consent.
Try it now
Prepare a file before uploading
Take a work file containing people's data, replace names with codes, and remove numbers. Ask yourself: could a stranger identify the person from this file? If not, upload it.