Chatbots are designed to make you feel like your conversations are personal. They’re highly agreeable, and over-friendly, and simulate the experience of chatting with a friend online—assuming that friend can instantly generate a wall of text in a few seconds. In reality, however, they’re far from personal, especially if your definition of “personal” includes “for your eyes only.”
Claude chats were visible on Google this weekend
Some Claude users can attest to that fact after the weekend. As Cyber Security News reports, Anthropic’s Claude share links have appeared in Google search results, indicating search engines like Google have indexed users’ chats. Over the weekend, a viral Reddit post highlighted how, by simply adding “site:claude.ai/share” to your search queries, you could pull up real Claude chats directly in Google.
If a Claude user talked about the keyword you searched, and Google indexed that conversation, you could pull it up and read the transcript. According to Cyber Security News, that included lawyers discussing legal strategies, engineers talking through technical issues, and users being open with their personal problems. Depending on the conversation, that could range from mildly embarrassing to downright mortifying.
Now, some caveats: Anthropic quickly nipped the issue in the bud after the Reddit thread started making waves. If you try to search anything with “site:claude.ai/share” now, you’ll simply receive a failed search result in return, so all your Claude chats should be safe going forward. Perhaps most importantly, it doesn’t seem likely that these indexed chats affected everyone who uses Claude; rather, these were “share links,” which Claude generates when you share a conversation with someone else. The glitch appears to stem from this feature—if you shared a chat from Claude, it may have ended up in this searchable database.
That said, not all the leaked chats might have fit that criteria. This same situation happened back in September, and, as Futurism reports, at least one affected user denied sharing their chats. Taken at their word, it’s possible that some exposed Claude chats were also from users who never used the share chat function.
AI chats aren’t private
Even if you’re confident your Claude chats weren’t shared, or trust that the leak has been plugged, that doesn’t mean that your AI chatbot conversations should be treated with an assumption of privacy. Even if you never shared a conversation from Claude, and thus likely weren’t affected by this issue, your chats may not be contained within your account. While you might think what you say with your AI stays between you and the chatbot, everything you say to your bot is shared with Anthropic and used to train future models—unless you explicitly opt out.
That’s not just Anthropic: Most, if not all, AI companies operate this way, and each has its own set of policies. Some may have humans directly reviewing your chats, for example—something you might not think about when asking your AI bot for help with a personal issue.
Even if you opt out of data training, the company in question may still hold onto your chats for a set period of time. Anthropic, for example, will hold onto your chat data for 30 days after you delete the conversation. So not only is that data held on Anthropic servers in the first place, it lives there after you can no longer access it. Plus, Anthropic isn’t the only company dealing with indexing exposure: While this is the second time this happened to Claude users, ChatGPT users dealt with a similar problem last August.
If you do choose to use chatbots, never share personal or private information with them—especially highly sensitive data like health info. AI is undoubtedly here to stay, and many find it useful, but if you care about your privacy, these tools simply don’t have your best interests in mind.
