Artificial Intelligence

Grok AI Chatbot Exposes 370,000+ User Conversations Online

AUG 22 2025·Updated AUG 23 2025

Story

Above: The Grok logo on a mobile phone. Image credit: Nikolas Kokovlis/NurPhoto/Getty Images

The Facts

Elon Musk's xAI chatbot Grok has made over 370,000 user conversations publicly searchable on Google through its share button feature, which creates unique URLs that are automatically indexed by search engines without clear user warnings.
BBC News
Forbes
The exposed conversations included sensitive personal information such as medical questions, passwords, names and personal details, as well as requests for instructions on manufacturing illegal drugs like fentanyl and methamphetamine.
CNET
Digital Watch Observatory
Some leaked conversations showed Grok providing detailed instructions on bomb construction, malware creation, suicide methods and even assassination plans targeting Elon Musk himself, violating xAI's own terms of service.
LBC
Forbes
The privacy breach occurred when users clicked Grok's share button to create shareable links, which were then automatically made available to search engines like Google, Bing and DuckDuckGo without explicit user consent.
Digital Watch Observatory
Verge
This incident follows a similar privacy breach with OpenAI's ChatGPT earlier this month, which OpenAI quickly addressed by discontinuing the feature after calling it a short-lived experiment that created accidental sharing risks.
Mashable
Forbes
Users can manage their exposed conversations by deleting their shared chat history, though it remains unclear if removal affects content already indexed by search engines.
MINT
CNET

Sources Split

LEFT

CENTER

RIGHT

ESTAB
CRIT

CENTER

PRO
ESTAB

The Spin

Techno-skeptic narrative

This massive privacy breach shows exactly why AI companies can't be trusted with personal data. Grok secretly published hundreds of thousands of private conversations without proper warnings, exposing everything from medical questions to passwords. AI companies clearly won't keep even the most sensitive data safe from exposure.

CNET

Prolificnorth

Techno-optimist narrative

The Grok situation highlights normal growing pains in AI development, similar to issues other major platforms have faced and resolved. Users who clicked share buttons were actively choosing to create shareable links, and cannot have the same expectation of privacy.

Forbes