Exposed: xAI’s Grok App Reveals Public Conversations

In a shocking incident, more than 370,000 private conversations from xAI's Grok app were inadvertently made public after a design flaw in its sharing feature allowed search engines to index the URLs of shared conversations.

The issue arose when users clicked the 'Share' button on Grok, which created public pages for conversations. Since these pages weren't access-controlled or flagged as "noindex," search crawlers were able to follow and index them, making ordinary chats and even attachments discoverable to anyone who searched for them.

This incident is not an isolated one. In July, ChatGPT's opt-in "discoverable" share links also ended up indexed, prompting OpenAI to disable the feature and coordinate removals. The Grok incident highlights the need for companies and individuals to prioritize data privacy when using AI tools.

Who is Most Exposed?

Anyone who has used AI tools for personal or work-related tasks could be at risk of having their conversations exposed. However, certain groups are more vulnerable to this type of incident:

  • Talking about sensitive topics: Individuals who use Grok or other AI apps to discuss confidential matters may have had their conversations exposed without their knowledge.
  • Business users: Employees using AI tools for work-related purposes may be at risk if their conversations are indexed and made public.
  • Past users of the app: Anyone who has used Grok in the past, even if they no longer use it, may still have their old conversations exposed if search engines find them.

Expert Advice on How to Fix the Issue

Anirudh Agarwal, CEO of OutreachX, emphasizes the importance of taking swift action to mitigate the damage. "A share link is a publication, not a whisper," he notes. "Once a crawler can reach it, you trigger distribution, not just disclosure; caches outlive your delete button." To protect yourself and others, follow these steps:

1) Check if Your Chats are Public (within 2 minutes)

Open an incognito window and search for the shared conversation URL. Repeat this process on Bing and DuckDuckGo, saving each URL you find. Reporters have verified that Grok share pages were being indexed in this way.

2) Delete the Conversation at the Source (inside X/Grok)

Follow these steps to delete a conversation from the source:

  • For web: Settings → Data controls → Shared links → Manage In the modal, click the trash icon to delete a shared link or the chat itself.
  • For Android: Tap the menu (≡) in the top-left. Locate the chat, press and hold the title. Tap the red Delete option.
  • For iOS: Tap the menu (≡) in the top-left. Find the chat, press and hold its title. Tap Delete (red).

3) Do the Same for Bing/DuckDuckGo

Submit the links via Bing Content Removal; since DuckDuckGo sources traditional links largely from Bing, this helps both.

Data Privacy vs. Chat Leaks: A Critical Review

The Grok incident highlights a critical issue in AI tool design: inadequate protection for sensitive data. A single design flaw can turn private conversations into public records, exposing users to significant risks.

This incident underscores the need for companies and individuals to prioritize data privacy when using AI tools. New EU AI Act obligations for general-purpose AI are now in effect, setting a new standard for privacy-respecting defaults in AI products.

Conclusion

The Grok incident serves as a reminder that even seemingly minor design flaws can have significant consequences. By taking proactive steps to secure our data and follow expert advice on how to fix the issue, we can minimize the risks associated with AI tool use.