Skip to content
cisonews.ai
Data exposure

Shared ChatGPT conversations found indexed on Google search in 2025

OpenAI pulled a discoverability option after shared ChatGPT links began turning up in Google search results, exposing sensitive conversation details.

Event
2025-08-05
Affected
OpenAI ChatGPT
Published
Sources
1

Compiled October 13, 2026 from public records. The events described took place in week of August 5, 2025.

By the cisonews.ai Desk · Published

OpenAI removed a feature that let ChatGPT users make shared conversation links discoverable by search engines, after thousands of those shared chats turned up in Google search results.

OpenAI's chief information security officer disclosed the removal in early August 2025, calling the discoverability option a short-lived experiment.

What happened

Users had to choose to make a shared ChatGPT conversation discoverable through a pop-up window before the link could be indexed by a search engine.

A media report found that once that option was enabled, Google crawled and indexed the resulting pages, surfacing thousands of ChatGPT conversations in ordinary search results.

Reporting also raised concerns that some people enabled sharing only intending to forward a link to friends or family, not realizing it could become publicly searchable.

Impact

The indexed conversations did not contain directly identifiable information, but several included specific details that could help someone identify the human participant in the exchange.

Unknown is the exact number of conversations indexed or whether any specific individuals were identified as a result.

Response

OpenAI said the feature introduced too many opportunities for people to accidentally share things they did not intend to, and removed the discoverability option.

The company also said it was working to remove already-indexed shared conversation content from affected search engines.

Lessons for security leaders

Opt-in sharing features for AI tools can behave like publishing tools once a link is indexed, so security teams should review any discoverability or public-link settings in AI products before employees use them. Clear in-product labeling is not sufficient on its own; organizations should assume that any shareable AI conversation link could become searchable and should set internal policy accordingly. Security teams should periodically search for their own organization's shared AI conversation links appearing in public search engines and request removal where sensitive data is exposed.

Sources (1)

  1. [1]
    Fortune
    fortune.comAugust 5, 2025tier B · Press

Spotted an error? Request a correction. Every fact on this page is tied to a listed source.