DeepSeek User Chats Exposed in Public Search Results

Private conversations between users and the Chinese AI model DeepSeek have become publicly accessible through a simple search query, according to reports from TASS. The discovery raises fresh concerns about data privacy and how shared content can resurface online.

When users enter a specific search command into Google, the results reveal full chains of dialogue, including document titles, user questions, and AI responses. However, downloading the actual files is not possible. MWS AI, a cybersecurity firm, confirmed that no new leaks of ChatGPT or DeepSeek conversations have occurred, and there are no reports of hacked services or unauthorized access to private exchanges.

The exposed dialogues appear to be from the Shared Conversations feature, where users voluntarily share their chats. Analysts estimate these conversations are roughly a year old. MWS AI stressed that searching cannot break into personal accounts or read private dialogues—only those intentionally shared are visible.

This incident follows warnings about fake DeepSeek websites that trick users into downloading malicious archives disguised as the AI client for Windows. Experts advise caution when sharing any AI conversation publicly, as even temporary sharing can lead to long-term exposure.