cd /news/ai-safety/claude-users-shared-conversations-we… · home topics ai-safety article
[ARTICLE · art-76708] src=fastcompany.com ↗ pub= topic=ai-safety verified=true sentiment=↓ negative

Claude users’ shared conversations were showing up in Google searches

Thousands of shared Claude conversations and artifacts were indexed by Google searches earlier this month, exposing potentially private data including sexting chats and proprietary information, cybersecurity analyst Rachel Tobac of SocialProof Security said. Anthropic has since delisted the results, but Tobac called it "the exact issue we talked about before," echoing a similar ChatGPT flaw reported by Fast Company in July 2025. An Anthropic spokesperson said the company does not share chat directories with search engines and that links become public only when users deliberately share them.

read2 min views1 publishedJul 28, 2026

Thousands of conversations between Claude and its users, alongside tools created during those chats, were available for a period of time through Google searches earlier this month—echoing a similar issue that blighted OpenAI’s ChatGPT and was first reported by *Fast Company *almost a year ago to the day.

Among the conversations available through a simple Google search that has now been fixed were purportedly those involving sexting and one where an engineer is launching “a project meant to train future Security Engineers in Web/Cloud/AI Security.” Redditors spent large parts of the weekend uncovering embarrassing or potentially private conversations, but Anthropic appears to have delisted the results from Google over the weekend.

“This is wild to see,” says Rachel Tobac, a cybersecurity analyst and CEO of SocialProof Security. “It’s the exact issue we talked about before.”

In July 2025, *Fast Company *exclusively revealed a similar issue with ChatGPT that OpenAI initially suggested was not significant because the information that chats shared when a checkbox was ticked could be indexed by search engines was presented in smaller, lighter text in a popup. It then pulled the ability to opt in to allowing chats to be indexed by search engines within a week.

Anthropic shows similar language to users, who appear not to understand the consequences of sharing published links. “This is an obvious user experience issue,” Tobac says. “Users clearly do not understand what they are agreeing to or that they are publishing their private chats with passwords, keys, personal questions, proprietary data, tax details, et cetera to the world.” Google also faced a similar issue in 2023.

As well as chats, some artifacts—or tools and websites—that users created with Claude that were chosen to be shared were indexed by search engines, including one supposedly comparing two treatment arms of a medical trial. Fast Company is choosing not to share the link to that artifact because it contains names, ages, and treatment dates, though it is not clear whether they are the names of real people in a real trial. Another contains what appears to be a database of access codes to Irish residential apartment blocks—which again, *Fast Company *is choosing not to share publicly.

“In this case, it’s essential that the Claude team in charge of this function make it more clear to users that when they click that publish button, that their chat will be publicly visible and indexed online—including on Google search,” Tobac says. “Or better yet, remove it until users understand and can make an informed decision that impacts their security and privacy.”

An Anthropic spokesperson tells *Fast Company: *“We give people control over sharing their Claude conversations publicly, and in keeping with our privacy principles, we do not share chat directories or site maps with search engines like Google.”

The spokesperson says the links cannot be found or predicted by outsiders. They become accessible only when a user deliberately shares one, at which point the conversation is effectively public and can be saved by third-party archiving services.

── more in #ai-safety 4 stories · sorted by recency
── more on @anthropic 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/claude-users-shared-…] indexed:0 read:2min 2026-07-28 ·