Anthropic’s Claude chatbot is facing renewed scrutiny after shared conversations and AI-generated projects briefly appeared in Google search results.
Reddit users discovered that typing search operators like “site:claude.ai/share” into Google surfaced a long list of shared Claude conversations and Artifacts, the interactive apps and documents users build inside the chatbot.
The exposed material ranged from programming notes and fake book reviews to far more sensitive content: patient medical reports, clinical trial data with real names attached, internal company documents, employee reviews, API keys and login credentials, according to Futurism and 404 Media.
The culprit was Claude’s “Share” button, which generates a public link so a conversation can be sent to a friend or colleague. Claude’s interface warns that “anyone with the link can view,” but nothing suggests those links could land in a search engine.
How the leak happened
Anthropic uses a robots.txt file to tell web crawlers to stay away from shared chat pages, and that instruction has been in place since at least September 2025, according to WIRED. But robots.txt isn’t foolproof.
Google and Bing’s developer guidance says crawlers can still index a blocked page if it’s linked to from somewhere else on the web — a forum post or social media share, for instance — unless the page also carries a “noindex” tag. WIRED reviewed exposed Claude pages and found no such tag present.
Anthropic’s response
Anthropic spokeswoman Amie Rotherham told TechCrunch: “We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or sitemaps with search engines like Google. These shareable links are not guessable or discoverable unless people choose to share them themselves. When someone shares a conversation, they are making that content publicly accessible, and like other public web content, it may be archived by third-party services.”
Google spokesperson Ned Adriance placed the responsibility on Anthropic, saying, “Neither Google nor any other search engine controls what pages are made public on the web, and these pages were indexed across many search engines. We give site owners clear controls to decide whether pages can be crawled or indexed, and we always respect those directives.”
By Monday, the specific search queries that surfaced the chats had stopped returning results, though individual links that had already been shared elsewhere remained live and reachable.
Not Claude’s first rodeo
Forbes reported a nearly identical problem last year, when Google had indexed several hundred Claude conversations. The same basic failure has now hit three major chatbots: roughly 100,000 ChatGPT conversations turned up in Google search last year, and Elon Musk’s Grok has been caught in similar territory.
Why this repeats
The pattern across OpenAI, Anthropic and xAI suggests this isn’t a one-off bug but a structural blind spot in how AI companies build sharing features.
Chat platforms borrowed the “shareable link” concept from tools like Google Docs without replicating the safeguards those tools rely on.
And unlike a shared document, a shared AI chat often contains a running, unfiltered record of what someone was thinking through in the moment, such as health worries, legal questions, financial details, making the stakes of a leak considerably higher than a stray spreadsheet.
What users should do
Anthropic advises checking Settings > Privacy > Shared Chats and revoking any links no longer needed. Until sharing tools carry the same protections as the rest of the internet, treating any “share” button as potentially public — not private — is the safer bet.
Also read: 1Password’s Claude sign-in integration lets the AI complete authenticated tasks without receiving passwords or one-time codes, while requiring user approval.