China’s AI Startup DeepSeek dominates the headline of the AI model and pays attention as well as technology. Deepseek-R1’s reasoning model is praised for being consistent with O1 of Openai while providing a more cost-effective approach.
With the greater momentum of DeepSeek, the security researchers of Wiz -based cyber security companies, based in New York, discovered a major security. On Wednesday, Wiz reported a mountain of confidential data from China’s AI Startup DeepSeek, which was exposed to the open Internet.
DeepSeek has released confidential information including chat history.
In a blog post, Wiz stated that DeepSeek’s infrastructure scan has revealed more than one million units of data. The file contains a digital software key and chat log, and the exchange of users with the company AI assistant has been captured. Wiz Research discovered an exposed DeepSeek database that leaked sensitive information, including chat history.
“The published database belonging to the Deepseek has completely controlled the database operation, such as the function of accessing internal data. Exposure contains more than 1 million logstreams with very sensitive information. I am, “says Wiz.
According to Wiz, DeepSeek immediately dealt with this issue. Ami Lutowak, the highest technical officer of Wiz, said that he acted quickly when Deep Shek received the notification.
“They defeated it within an hour,” Luttwak said. “But it was very easy to find what we found not only me.”
On the other hand, DeepSeek has not yet commented on this case.
The company is at the center of the rapid increase in the AI adoption, drawing out the excitement in China, drawing concerns in the United States, and the AI Assistant is comparable to Openai’s technology at one minute, so the rise of DeepSeek is the rise. We promoted questions about long -term sustainability. American AI company. By Monday, the download of Apple’s App Store overtook ChatGpt and caused a wider high -tech stock.
This latest security slip emphasizes the growing reality. AI’s breakthrough brings more than just competitive. Miss steps may occur as companies compete to push the boundary. In this case, the problem was quickly corrected, but there is a bigger question. How many times do you notice the data as the AI expands?
Source link
