Jin Daily AI Trivia – OpenAI Accidentally Leaks User Prompts to Google

Jin Daily AI Trivia – OpenAI Accidentally Leaks User Prompts to Google

Instead of paying Google for proper API access to search data, ChatGPT was doing something way sketchier: it literally pasted your entire conversation into Google’s search URL to scrape results.

And here’s the kicker — webmasters ranking high enough for certain keywords can now see all those ChatGPT prompts in their Google Search Console. Yup, your “private” AI chats? Not so private anymore.

The bug was discovered after ChatGPT accidentally appended extra URL parameters (like https://openai.com/index/chatgpt/) to its searches, causing a flood of random user prompts to show up in webmasters’ search performance reports. (because their blog rank high for that URL)

Think about it: someone types “what should I do, I accidentally deleted my company X’s AWS server” — and now some random website owner who ranks for AWS troubleshooting can see that exact query in their analytics dashboard.

ChatGPT Altas browser made it even worse by automatically searching everything users typed, turning every single prompt into a potential data leak.

So much for local LLMs and privacy, right?

Bad for users, goldmine for webmasters — but definitely not what anyone signed up for when they thought their AI chats were confidential.