You find a brilliant workaround, a perfect prompt, or maybe you just vented to an AI about your boss. You click “Share” to send the link to a coworker. You think it’s semi-private. It’s not. It’s entirely public, indexed by Google and Bing, and the whole world can read it.
The “share” button wasn’t built for your friends; it was built for the search engine crawlers.
Recently, a user on Reddit discovered exactly this. If you search specific operators on Google or Bing, you can pull up thousands of Claude shared conversations. Some are mundane. But others? They contain sensitive business logic, personal confessions, and proprietary code. You trusted the walled garden. The walled garden had a backdoor wide open to the web.
But here is where it gets genuinely embarrassing for Anthropic. Among the exposed chats, many are labeled “shared by Anthropic.” That’s right. The company building the “safe” AI accidentally indexed its own internal demos and employee conversations. It’s a classic dogfooding failure. They built the trap, stepped in it first, and didn’t bother to check if the rest of us were falling in alongside them.
Security theater is when a company locks the front door but leaves the windows wide open for the search bots to crawl in.
This isn’t just a technical glitch; it’s a breach of trust. When you click “share” on a conversation, there is an implicit contract. You believe you are generating a unique, obscure URL for a specific person. You don’t expect it to become a permanent, searchable record of your intellectual property or personal anxieties. Yet, that’s exactly what happened. Bing is even throwing up Cloudflare captchas to hide their tracks, but the data is there, waiting to be scraped.
Privacy isn’t a feature you can patch in later; it’s the foundation you build the house on.
Stop sharing your Claude conversations right now. Treat every AI chat box like a public forum. Because until Anthropic fundamentally fixes how these links are generated and indexed, your secrets are just a Google search away. The age of AI confidants is over. Welcome to the age of AI broadcasting.
FAQ
Q: Isn't any shared link technically public anyway?
A: No. Generating a share link implies a direct, obscure URL meant for a specific recipient, not an open invitation for Google to crawl, index, and rank your private conversations for the world to find.
Q: What's the practical implication?
A: Stop using the share feature immediately. You must assume any sensitive data, proprietary code, or personal information you put into an AI chat can and will be exposed to the public internet.
Q: What's the contrarian take?
A: This isn't a bug, it's a feature. AI companies want your prompts and interactions to train future models, and what better way to harvest high-quality data than by making you voluntarily index it for them?