So… this is awkward, maybe even a little embarrassing— but people are unknowingly publishing their ChatGPT convos to Google.

Not even kidding.

If you search site: chatgpt.com/share, you’ll find actual conversations users had with ChatGPT—live, public, and just a little too real.

Some are wholesome (think: recipes and physics questions). Other are… absolutely unhinged (think: résumé rewrites that didn’t work and people trying to microwave forks).

One convo even ends with ChatGPT writing a full guide titled: “How to Use a Microwave Without Summoning Satan.” So yeah. That’s where we’re at.

But Wait—how are these chats ending up online?

Here’s the thing: ChatGPT doesn’t make conversations public by default.

To share a chat, a user has to:

  • Click the “share” button

  • Then click a second “create link” button

  • Then decide whether or not to make it discoverable

Problem is: If you made the link public—even by accident—Google or any other search engine can index it. And that means anyone can stumble on it.

Worse still: If you dropped names, company info, job titles, or locations into the chat? That’s all potentially traceable. And yes, someone already got doxxed by their résumé. (Their LinkedIn was just a few clicks away.)

And OpenAI’s response? “It was an experiment.”

Apparently, they were testing ways to make shared chats easier to find. The test might be over—but those links? Still up. Still very searchable.

Google, of course, says it’s just doing what search engines do: indexing public stuff.

Fair play. But yeah… that still leaves a bunch of shared convos floating around with no clear expiration date.

And this isn’t just a OpenAI problem

  • Meta’s AI was caught showing private chats in public feeds

  • Google’s Gemini had /share links popping up in search results

  • Plus as some pointed out, ChatGPT’s answers are starting to look suspiciously similar to Google’s AI overviews

So no, this isn’t some one-off bug. It’s part of a much bigger data hygiene and exposure problem.

And here’s what you should do (like right now)

If you’ve ever shared a ChatGPT link—even casually—go check what’s out there.

  • Search your company name + site with: chatgpt.com/share

  • Audit any shared convos (especially ones with names or internal info)

  • Delete anything that shouldn’t be public

  • Talk to your team about how AI outputs are handled

  • If privacy matters, consider AI tools with private cloud or on-prem setups

Bonus tip: If you’re sharing AI-generated content with anyone outside your org, treat it like a public document. Because it might be.

One last thing: To delete your shared ChatGPT links, head to: Settings → Data Controls → Shared Links → Manage

In short: This wasn’t a bug. It wasn’t a leak. It was a feature—just one most people didn’t realize could turn a private convo into a public file.

So yeah. Might be a good time to check if you’ve accidentally left anything out there.

Here’s the full report.

Reply

or to participate

More From The Automated

No posts found