Can You Use ChatGPT for Confidential or Privileged Documents?

If you handle other people’s sensitive information for a living — a lawyer with privileged files, a therapist with session notes, an accountant with client books — you’ve probably had this thought: “AI could do this in seconds. But can I actually put this document into it?”

It’s the right question to ask. And the honest answer is that with a cloud tool like ChatGPT, you should be very cautious. Here’s why, in plain terms.

What happens when you paste a document into ChatGPT

When you upload a file or paste text into ChatGPT, three things happen that matter for confidential work:

1. It leaves your computer. Your text is sent over the internet to OpenAI’s servers, where the model actually runs. It doesn’t process anything on your machine.
2. It’s stored under your account. Your conversations are retained and tied to your identity. Depending on your plan and settings, they may be reviewed by staff for safety or abuse monitoring.
3. It may be used to train future models. On consumer plans, your inputs can be used to improve the model unless you’ve turned that off — and even then, the data has still traveled to and been stored on a third party’s servers.

None of that makes ChatGPT “insecure.” Cloud providers invest heavily in security, and for everyday, non-sensitive use it’s perfectly reasonable. The problem is narrower and more important: once the data leaves your machine, you no longer control it, and you can no longer promise your client that it never went anywhere.

Why “just don’t put anything sensitive in it” doesn’t work

The usual advice is to de-identify everything first, or simply avoid pasting sensitive material. That sounds sensible until you try to do the actual work.

• A progress note is the confidential information — there’s nothing to strip out.
• A contract you want summarized only matters because of the specific names, numbers, and terms in it.
• A discovery document dump loses its meaning the moment you redact what makes it discoverable.

For professionals with a duty of confidentiality, “be careful what you type” isn’t a workflow. It rules out the very tasks where AI would help most.

The one approach that removes the risk entirely

There’s a simple principle that cuts through all of this: the only data that can’t leak is data that never leaves your device.

That’s the idea behind running AI locally. Instead of sending your text to a company’s servers, the AI model runs directly on your own computer. Your documents are processed on your machine and are never uploaded. There’s no account tying the work to you, and nothing is transmitted that could be intercepted, stored, or subpoenaed.

You can even prove it to yourself: disconnect from the internet, and a local AI keeps working. There’s no server for your words to travel to.

This is exactly what WorkInPrivate is built for. It’s a desktop app that runs an open-source AI model on your own computer — so you get ChatGPT-style help with writing, summarizing, and reading documents, without any of it leaving your machine. No cloud, no account, no telemetry.

Where this matters most

Different professions run into the same wall from different angles. We’ve written more specific guides for several:

• Private AI for lawyers — work through privileged documents without third-party servers
• Private AI for therapists — draft and summarize notes that never leave your laptop
• Private AI for accountants — analyze client financials that stay on your machine
• Private AI for healthcare — summarize and draft with patient information kept offline

The bottom line

Can you use ChatGPT for confidential or privileged documents? Technically yes — but it means trusting that a third party’s servers, staff, and settings all behave the way you hope, and accepting that the data has left your control. For work you’re ethically or legally bound to protect, that’s a hard promise to keep.

Running the AI on your own device sidesteps the whole question. If you want to see the difference laid out side by side, here’s how a private ChatGPT alternative compares, and a deeper look at whether ChatGPT is really private.

WorkInPrivate keeps your data on your device, which removes the third-party-disclosure risk of cloud AI. It supports your confidentiality obligations — it isn’t a substitute for your own compliance program.