Before You Tell AI Something Personal, Read This First

The Simple Habit I Started Using Before I Paste Anything Into ChatGPT or Claude

I was about to paste a work email into ChatGPT, the whole thing, copied and dropped in without a second thought. I wanted help figuring out how to respond to something a little tricky. My cursor was hovering over send when I stopped and actually looked at what I'd just pasted — names, a coworker's title, details about a situation that weren't really mine to hand over so casually. None of it was secret. None of it felt especially sensitive. It just wasn't necessary, and I'd included it out of habit, not because the help I needed actually required it.

That pause turned into something I do now almost automatically, whether I'm dealing with a work email, a school document, medical paperwork, or a contract I don't fully understand. Before I share anything, I ask myself a simple question: does it actually need this detail to help me? Most of the time, the honest answer is no.

Why I Used to Paste Everything

For a while, I didn't think twice about it. If I needed help with an email, I pasted the whole email. If I needed help understanding a contract, I pasted the whole contract. If I wanted feedback on a résumé, I pasted my full résumé, contact information and all, without stopping to consider whether any of that actually mattered to the help I was asking for.

I think most people do this the same way I did — not carelessly, exactly, just automatically. It feels natural to hand over the complete picture, the same way you'd forward an entire email to a friend for advice rather than summarizing the relevant part. It doesn't occur to you to ask whether the names, account numbers, or identifying details buried inside actually change the answer you're going to get.

What Changed

What shifted for me was realizing that AI usually cares about the problem, not the identity of the people involved in it. The help I need with a tricky email doesn't depend on knowing my coworker's actual name. The help I need understanding a contract clause doesn't depend on which company issued it. The advice on a medical question doesn't require knowing which doctor said what.

So instead of "my son Wyatt," I've started saying "my middle-school son." Instead of naming my employer, I just describe the situation generally. Instead of identifying a specific doctor, I describe the concern itself. Instead of pasting an entire contract, I share only the clause I'm confused about after removing names, account numbers, and other identifying details. I've realized I can usually describe the situation without describing the person. The advice doesn't change just because I leave someone's name out of it.

This isn't about hiding information or being suspicious of the tool. It's about minimizing what I hand over to only what's actually useful for solving the problem in front of me — which, it turns out, is almost never the identifying details I used to include automatically.

Where AI Actually Helped

What surprised me most is how little the quality of help changed once I started leaving identifying details out. Explaining an unfamiliar medical term doesn't require my actual diagnosis attached to my name. Reviewing a résumé for clarity and structure doesn't require my home address or phone number sitting at the top of it. Summarizing a dense contract doesn't require knowing which specific company or landlord wrote it.

I've used it to organize school information without including my son's full name or his school. I've used it to make sense of insurance paperwork by describing the coverage question, not by uploading a document with policy numbers and personal identifiers attached. I've had emails rewritten by describing the situation and the tone I'm going for, rather than pasting the original with everyone's names still in it. I've even compared job offers by describing the terms — salary range, benefits, schedule — without naming the actual companies involved.

In almost every case, removing the identifying details changed nothing about how useful the response was. The help was there. I just stopped handing over more than the help actually required.

What AI Couldn't Know

There's a limit to how far this habit can go on its own, and it's worth being honest about it. AI has no way of knowing whether something is covered by a confidentiality policy at my job, whether a client relationship comes with obligations I need to respect, or whether an NDA I signed restricts what I'm allowed to share about a project. It doesn't know the specifics of HIPAA or FERPA as they might apply to a particular piece of information I'm holding, and it can't tell me whether something technically belongs to someone else — an employer, a client, a school — in a way that changes whether I should be sharing it at all.

It also doesn't know what policies my workplace, school, or organization expects me to follow. None of that is something AI can decide for me. That responsibility — knowing what I'm actually allowed to share and choosing to share less than everything by default — stays entirely mine.

The Prompt I Use Before Sharing Information

When I'm about to paste something in and I'm not sure how much detail is actually necessary, this is what I use to keep myself honest about it.

I want help with the following situation: [DESCRIBE YOUR PROBLEM] Before asking me to upload documents or provide additional details, tell me only the information you actually need to help me. If I start including information that probably isn't necessary—such as names, addresses, account numbers, passwords, confidential work information, medical identifiers, or other personal details—ask me to replace those with generic descriptions instead. Help me solve the problem while sharing the minimum amount of personal information necessary.

A Small Habit That Changed How I Share

None of this comes from a place of thinking AI is dangerous. It's not that dramatic. It's just that I've become more intentional about a habit I used to do without thinking at all.

Now I pause before I paste. I don't ask myself, "Is it okay to share this?"

I ask,

"Does AI actually need this detail to help me?"

Most of the time, the answer is no. That small habit hasn't made AI less useful. It's made me more intentional about what I share—not just with AI, but anywhere online.