Some private conversations with Anthropic Claude are now visible in Google and Bing search results. Users recently found that threads containing political advice, legal queries about Kansas attorneys, and erotic role play were indexed by major search engines.
Claude lets people create public links to specific chat threads. These links function as standard web pages, which means search engines can crawl them unless specific instructions are followed.
Anthropic tells web crawlers to ignore shared chats via a robots.txt file. The company has maintained this restriction since at least September 2025, according to records on the Wayback Machine.
However, preventing pages from appearing in search results is more complex than simply blocking crawlers. Bing documentation notes that developers should use robots.txt alongside a noindex tag on individual pages. Bing currently displays about 612 results for the query site:claude.ai/share.
Google states it ignores robots.txt instructions if a page is linked to from elsewhere online and lacks a noindex tag or an x-robots-tag in the response header. WIRED checked the exposed Claude pages and found they were missing the noindex tag that both search engines consider.
Several chats flagged in a Reddit post were deleted by the time the story was reviewed, and some results no longer appear in Google searches. Microsoft and Anthropic did not comment on the issue.
Ned Adriance, a Google spokesperson, said indexing these chats is Anthropic’s responsibility. He added that Google respects directives given by site owners but does not control what pages are made public on the web.
Anthropic has faced similar criticism last September. It told Forbes it uses robots.txt to stop crawlers from accessing shared chats, but the report noted there is no guarantee this stops indexing.
AI labs still rely on robots.txt for other reasons. Many promise creators their sites will not be used for training data if developers disallow certain crawlers. Anthropic, Meta, and OpenAI all include instructions in their robots.txt files to block competitors’ crawlers from their chatbot sites.
OpenAI and Meta did not respond to requests about blocking Google’s training crawler, Google-Extended. Google did not address questions about its competitors blocking its own crawler.
Although shared chats are not currently showing in Google results, the pages reviewed still lack a noindex tag. They could appear in search engines again. Users can manage access to keep conversations private via Settings > Privacy > Shared chats.
What it means
Users relying on public links to share Claude conversations should assume those links are not private. The current setup depends on search engines respecting robots.txt, which is not guaranteed. Adding a noindex tag would be the only way to ensure these specific pages remain out of search results.
