Skip to content

xAI

Disclosed Aug 18, 20251 year agoUnverified

Grok website exposed system prompts for its AI personas

The Grok website exposed the system prompts for several of xAI's AI personas, including a conspiracy theorist and an unhinged comedian; TechCrunch confirmed the exposure first reported by 404 Media.

What is known

People affectedNot stated in the sources we have
DisclosedAug 18, 2025
AttackAI or model
Data exposedPrompts and chats, Messages, Internal documents
SectorAI · US
StatusUnverified: not yet confirmed by an official notice, a filing or the organization

Sources

Notices filed

WhereFiledPeople
ResearchtotalAug 18, 2025
ResearchtotalAug 20, 2025

Other breaches at xAI

History of this record
  • 2026-09-25 · data_types: ["prompts","messages"] to ["prompts","messages","internal-docs"] · seed source
  • 2026-09-25 · disclosed: 2025-08-20 to 2025-08-18 · seed source
  • 2026-09-25 · summary: Forbes reported that URLs created when Grok users shared conversations were indexed by Google, Bing and DuckDuckGo, making hundreds of thousands of chats publicly searchable. to The Grok website exposed the system prompts for several of xAI's AI personas, including a conspiracy theorist and an unhinged comedian; TechCrunch confirmed the exposure first reported by 404 Media. · seed source
  • 2026-09-25 · title: Hundreds of thousands of Grok conversations searchable on Google to Grok website exposed system prompts for its AI personas · seed source
  • 2026-09-25 · added · seed source

First seen 2026-09-25 (Research). Record counts are as reported. Not legal advice.

Everything about xAI

New breaches by email

Wednesdays, only in weeks with newly disclosed breaches, the largest first.

Double opt-in. Unsubscribe any time.