Elon Musk’s Grok Chatbot Leaks

Updated:August 22, 2025

Reading Time: 2 minutes
Grok's chatbot

Something odd happened at Elon Musk’s xAI. 

It seems Grok, the chatbot, had a “share” button that didn’t stay private. 

Instead, over 370,000 conversations accidentally appeared on Google, Bing, and DuckDuckGo searches, shared Forbes

Imagine typing something personal, hitting share, and bam! Anyone could stumble on it.

How the Sharing Slip-Up Happened

Let’s break it down:

  • You click the “share” button, expecting to copy a link for friends or yourself.
  • That link suddenly turns into a public web page.
  • Search engines quickly index it. Now, anyone can find your chat with the right search.

One user said chats meant just for a team ended up on Google. “I was surprised… despite no warnings,” he shared.

What’s in Those Chats?

A lot of what Grok users asked for ended up online, too:

Shared TopicsWhy It Matters
Making fentanyl or methDangerous tips made public
Bomb buildingDetails of illegal activity in plain view
Malware and hacking toolsPotential for cyber threats
Suicide methodsHighly sensitive and alarming
Assassination plans (even of Musk!)Straight-up scary content
Personal info (passwords, docs)Real privacy risk

Everything from bomb recipes to deeply personal info like passwords made the rounds.

Didn’t OpenAI Mess Up Too?

Yes, OpenAI had a similar misstep just weeks earlier.

ChatGPT’s share feature created links that got indexed, too. OpenAI quietly shut it down, calling it a “short-lived experiment”.

Sadly, Grok didn’t follow that example fast enough. It didn’t warn users, didn’t clarify that it was public, and didn’t prevent it from happening.

Why This Matters

Privacy is a Two-Way Street

One tap, and your private chat becomes a public page. That’s scary, right? Especially if you share something meant for your eyes only.

Tech Moves Fast, but Users Don’t Always Catch Up

New features can be exciting. But without clear warnings, they can trip us up in big ways or expose us to real risk.

Mistakes Left Unfixed Can Spread

OpenAI fixed theirs fast. Grok’s version stayed live. It’s a reminder that delays in fixing errors can hurt everyone.

What Can You Do?

  • Think twice before sharing. If it’s sensitive, don’t tap that button.
  • Check your settings. See if Grok lets you limit visibility.
  • Ask for clarity.
  • Use trustworthy platforms for sensitive chats and know how sharing works.

A Note of Perspective

It’s tempting to blame bots or those running them.

But this feels more like a design shortcut. A button added without thinking, a tiny nudge that led to a big privacy stumble.

We’ve all clicked a “share” button without reading the fine print.

Now, those buttons need to be clearer. Until they are, it’s smart to treat every share with caution.

Before you hit “share,” imagine a stranger reading your chat. That quick mental check can save you a lot of worry, and protect what’s yours.

Onome

Contributor & AI Expert