On July 21, I identified a critical privacy breach in the DeepSeek ecosystem. SEO specialist David Konitsny discovered that Google had begun indexing public links to dialogues created by users via the "Share" feature. This is not just a technical glitch—it is a direct threat to privacy.
What ended up in search results?
Google's index included not just random conversations. It involved work documents, financial analytics, and personal correspondence, including Russian-language dialogues. According to my data, the number of indexed pages is in the thousands, and this is just the tip of the iceberg. Users, unaware of this, shared sensitive information that is now accessible to anyone who knows how to search.
How did this happen?
The "Share" feature in DeepSeek generates unique URLs that are not protected from search engine indexing by default. Google, as the most powerful search engine, crawled these links, and they appeared in search results. This is not a hack—it is a flaw in the platform's security design. Such links can only be removed manually through account settings, but most users are not even aware of this threat.
Analysis and consequences
I assess this incident as a serious failure in data management. DeepSeek, which positions itself as an innovative AI tool, should have anticipated protection from indexing by default. For users, this means their private data—from business strategies to personal conversations—has become publicly accessible. In my professional experience, such leaks often lead to reputational damage and even legal risks, especially if the dialogues contain confidential information.
My verdict: This is not just a technical error—it is a signal that security in AI services is lacking. Users should immediately check their account settings and revoke access to any published links. DeepSeek urgently needs to implement automatic indexing blocking for all shared dialogues, otherwise the reputational damage will become irreversible.