On July 25, a serious vulnerability was discovered on Reddit in the mechanism for publishing dialogues with Anthropic's AI assistant Claude. A search query site:claude.ai/share in Google revealed hundreds of other users' conversations, including seed phrases from crypto wallets, API keys, resumes with contact information, and numbers resembling U.S. Social Security numbers. This is not just a technical glitch—it is a direct threat to user privacy and asset security.

Only those dialogues that users published via the "Share" function ended up in search results. This function creates a link with a note that anyone with the address can see the chat, but it does not warn that the page may be indexed by search engines. Unlike "hidden" videos on YouTube, which use the noindex meta tag, Claude did not have one. Every published address automatically became a regular indexed page. Anthropic's robots.txt file was configured correctly, but this proved insufficient: Google indexes pages if there are links to them from open sources, even if the robot is denied access.

Why This Is Critical for the Crypto Community

In addition to seed phrases and access keys, the indexed chats contained lawyers' questions about professional ethics, political disputes, and erotic texts. But for us, as crypto market analysts, the key issue is the leak of seed phrases. This means attackers could have gained full control over wallets without leaving a trace. By July 26, Anthropic added noindex and nofollow tags, and Google began removing pages from search results, but some data had already spread through copies. A repository called Shared-Claude-Chats appeared on GitHub, containing an archive of 453 conversations with Claude and 519 dialogues with Grok—a total of 11,241 text messages.

A similar situation occurred with OpenAI in August 2025, when the "Make this chat searchable" feature sent thousands of dialogues to Google, Bing, and DuckDuckGo. Previously, Google also indexed user dialogues with DeepSeek, including work documents and personal conversations in Russian. This is a systemic issue: AI platforms do not fully understand the risks of automatic indexing of public links.

My expert opinion: The Claude incident is yet another reminder that any publication of data on the internet, even via a "secure" link, can be indexed. For cryptocurrency holders, this is especially dangerous: seed phrases and access keys should never appear in any chats, even if you trust the AI assistant. I recommend immediately checking your privacy settings in Claude and revoking access to published dialogues. If you have shared seed phrases—change your wallet. It is better to be safe than to lose your assets.