Claude’s Private Conversations Revealed in Google and Bing Search Results

What happens in between You and the AI ​​chatbot are not between you and the chatbot. Just ask Google.

At the end of the week, people were surprised to find out where others were Anthropic Claude chats can be easily found through internet searches. This story, which seems to have been first given by a redditora public chat that had people asking for advice on which political party to join, whether Kansas attorneys should self-describe if they believe they’ve broken the law, and a sex scene.

Claude they allow users to share chat “snapshots” with others by creating a public link to another chatbot thread. Some of the reasons these URLs are indexed by major search engines come down to the basic functions of websites, search engines, and the collision of the two when artificial intelligence enters the mix.

Basically, Anthropic advises web browsers, such as those used by search engines like Google and Bing, not to record chats that the user chooses to share with other people. The company does this through a file called robots.txt, which for a long time has been considered the most popular way to determine who has access to a website. Anthropic’s robots.txt has made “shared” chats off-limits to internet users since at least September 2025, according to a a snapshot on the Wayback Machine.

But it’s harder to prevent sites from being included in search results than you might expect.

Bing, which still shows “about 612 results” when you search for “site:claude.ai/share” at the time of this writing, it says in its technical documentation that developers can afford it log search engines use robots.txt—and web developers should too including “noindex” tag on every page.

Some of the chats that appeared in the results posted in Reddit were removed when WIRED saw them, and the results no longer appear in Google when you search for “share” the query still works on Bing. In a design guideGoogle says that it ignores robots.txt instructions if the site is linked to anywhere else on the Internet and the owner of the site does not include the special “noindex” html tag on the site or the “x-robots-tag” in the header of the site’s response.

WIRED reviewed a sample of Claude’s social media pages that appeared and found that they did not include the “noindex” tag that Bing and Google are said to consider when choosing whether or not to index a website.

Microsoft, which owns Bing, did not comment ahead of publication. Anthropic did not respond to multiple requests for comment.

Google spokesman Ned Adriance tells WIRED that Claude’s chat browsing is Anthropic’s responsibility. “Neither Google nor any other search engine controls which pages are displayed on the Internet, and these pages are displayed in most searches,”
Adriance says. “We provide the property owners logical controls to decide whether pages can be crawled or indexed, and we always respect that advice.”

Anthropic did not respond to questions about why it did not include “noindex” on its shared sites.

Last September, Anthropic was fired with the same story, and told Forbes that it uses robots.txt to inform crawlers that they should not access the chats they share. But as the Forbes report points out, there is no guarantee that it will stop search engines from indexing certain pages.

Although robots.txt doesn’t always work to prevent pages from being crawled by search engines, AI labs still use it for some reason. Many laboratories promise creators that their websites can’t be used as fodder for AI training as long as developers make sure to “disallow” other crawlers in their robots.txt files, and the labs follow the guidelines as well.