Wiz Research cybersecurity experts have detected a leak of confidential information from the DeepSeek database.

The Chinese neural network DeepSeek had confidential user data and system information publicly accessible.

This information is reported on the website of the American cybersecurity firm Wiz.

According to the company, anyone could view users' chat histories, authentication API tokens, secret keys, as well as sensitive system logs, unencrypted messages, and even access open passwords, download files from local devices, and gain access to server data.

“This database contained a significant amount of chat history, internal data, and confidential information, including log streams, API secrets, and operational data,” the statement said.

The company noted that its programmers were able to detect the leak “within minutes” and took action because “DeepSeek had caused a stir in the artificial intelligence field.”

As researchers from Wiz explained, they alerted DeepSeek to the issue through all available channels, but received no response. However, access to the database was closed within half an hour.

Background. Previously, Mind reported that American companies Microsoft and OpenAI are investigating the potential unauthorized acquisition of data related to technologies from the ChatGPT developer by a group linked to the Chinese AI startup DeepSeek.