As Google makes Gemini 2.0 free to all via the Gemini API in Google AI Studio and Vertex AI, cybersecurity experts have found code hidden deep in DeepSeek’s application that has the capability to send user data directly to the Chinese government.

The possibility of a direct path to the People’s Republic of China hidden in its code is very real, says Ivan Tsarynny, CEO of Feroot, an Ontario-based cybersecurity firm.

Under Tsarynny’s guidance, Feroot has developed proprietary technology that detects third-party scripts and their level of access to data that has been entered by users on the web — such as on registration and payment pages, as well as on website logins.

“We see direct links to servers and to companies in China that are under control of the Chinese government. And this is something that we have never seen in the past,” Tsarynny told NBC News.

Advertisers are looking for a less expensive AI model to train and then create advertisements, and to use it to run a portion of their company’s business models, including customer service. DeepSeek may not be the answer, although the industry is looking to reduce costs.

I cannot imagine a company using DeepSeek to support its customer service, considering that the app has a direct link to send data to China’s government, despite it being housed in the United States. And what about performance metrics, sharing performance metrics for any company with China’s government.  

DeepSeek’s AI Assistant, powered by DeepSeek-V3, overtook rivals such as Gemini and ChatGPT in late January to become the top-rated free application available on Apple’s App Store in the United States.

Cost is not the indicator of the quality of the model, an analyst told MediaPost. The price of a model does not directly correlate with its performance or effectiveness, and a high-quality model can sometimes be obtained at a lower cost. At what cost? OEMs began manufacturing products in China during the late 1990s or early 2000s. For some the quality declined. Is there a catch? 

Here’s the catch that Tsarynny made public. Registered users logging in to DeepSeek may unknowingly create accounts held in China, making their identities, search queries, and online behavior visible to Chinese state systems, according to the report.

Tsarynny used AI software to decrypt portions of DeepSeek’s code and found intentionally hidden code that has the capability to send user data to CMPassport.com, the online registry for China Mobile, a telecommunications company owned and operated by the Chinese government.

China Mobile was banned from operating in the U.S. by the FCC in 2019 because concerns of “unauthorized” access to customer data.

U.S. Reps. Darin LaHood, R-Ill., and Josh Gottheimer, D-N.J. have introduced legislation to ban the AI model DeepSeek on government devices based on a threat to national security, saying the company’s technology presents an espionage risk.

“The technology race with the Chinese Communist Party (CCP) is not one the United States can afford to lose,” LaHood said in a statement. “The national security threat that DeepSeek — a CCP-affiliated company — poses to the United States is alarming.”

The U.S. Navy banned DeepSeek in January.

Code hidden in DeepSeek’s application has been found with the capability to send user data directly to the Chinese government.