Exposed Data from Chinese AI Startup DeepSeek

By Raphael Satter
WASHINGTON (Reuters) – New York-based cybersecurity firm Wiz has discovered a significant amount of sensitive data from the Chinese AI startup DeepSeek that was unintentionally exposed online.

In a blog post released on Wednesday, Wiz revealed that scans of DeepSeek’s systems indicated the company had left over a million lines of unsecured data accessible. This information included digital software keys and user chat logs captured by the company’s free AI assistant.

According to Wiz’s Chief Technology Officer, DeepSeek acted quickly to secure the exposed data after being notified by Wiz. “They took it down in less than an hour,” said Ami Luttwak. “However, this was so simple to find that we believe we’re not the only ones who found it.”

DeepSeek had not responded to requests for further comment at the time of reporting.

The rapid success of DeepSeek’s AI assistant has garnered excitement in China while raising concerns in the U.S. The company’s capability to compete with industry giants like OpenAI at a lower price has led to questions about the viability of the business models of U.S. AI firms such as Nvidia and Microsoft.

As of Monday, DeepSeek surpassed U.S. rival ChatGPT in app downloads from Apple’s App Store, which has contributed to a decline in tech stock values globally.

(This story has been updated to clarify that Wiz is based in New York, not Israel, in the headline and first paragraph.)