South Korea's data protection authority has revealed that Chinese artificial intelligence app DeepSeek was transferring user data to overseas servers, including those in China, without user consent or proper disclosure.
The Personal Information Protection Commission (PIPC) said on Thursday that data such as device details, network information, and user inputs from AI prompts were sent to Volcano Engine, a Beijing-based cloud platform owned by ByteDance, the parent company of TikTok.
DeepSeek’s AI chatbot, R1, made headlines in January for delivering capabilities comparable to leading Western AI models at a significantly lower cost. But the app has since come under scrutiny in several countries, including South Korea, Italy, Australia, and some US states, over concerns related to data privacy.
The PIPC launched its investigation in February and has since halted the app’s availability in South Korea pending a thorough review of its data collection and handling practices.
"Initially, DeepSeek transferred personal data to companies located in China and the United States without obtaining users' consent or disclosing this in the privacy policy at the time the service was launched," said Nam Seok, a PIPC official.
He added that user inputs entered into the AI prompts were also among the data sent to Volcano Engine, raising serious privacy concerns.
Following the investigation, DeepSeek admitted it had not adequately considered South Korea’s data protection laws. It has agreed to cooperate with regulators and has voluntarily suspended downloads from Korean app markets.
DeepSeek has not responded to recent requests for comment but has previously stated that data was stored on "secure servers located in the People's Republic of China." The company claims its AI system is cost-effective due to its use of less advanced hardware.
In defense of the data transfers, DeepSeek reportedly said the data was used to "address security vulnerabilities and improve user interface and experience," according to the PIPC.
The app, which saw tens of millions of downloads shortly after launch, remains under tight scrutiny as regulators across multiple countries weigh the privacy risks posed by foreign-developed AI technologies.
