Taiwan: Authorities should not use DeepSeek for security reasons

DeepSeek's hyped large R1 language model and the chatbot based on it are not only causing tremors on the stock markets, but are also causing concern among numerous governments. According to consistent agency reports, Taiwan's Ministry of Digital Affairs declared on Friday that all government agencies and critical infrastructures in the island state should not use the AI system. The reason: it "jeopardizes national information security".

The DeepSeek service is a Chinese product, the department explained. Its operation involves cross-border data transmission and information leaks. There are also serious concerns about IT security and the protection of user privacy with the service at the latest after a leak. In principle, official bodies in Taiwan have been banned from using information and communication technology products and services that pose a threat to national information security since 2019. The government in Taipei has long accused Beijing of hybrid attacks, including cyberattacks and propaganda campaigns, which fall just below the threshold of open hostilities.

Countries such as Germany, Italy, South Korea, Ireland, France, Australia and the USA are also increasingly skeptical about DeepSeek. The Italian data protection authority, the Garante, ordered a national ban on the chatbot on Thursday. It urged the Chinese companies Hangzhou and Beijing DeepSeek, which operate the service, to stop processing data from Italian users with immediate effect.

Even keystrokes can be analyzed

The guarantor justifies the island order with the fact that DeepSeek had answered data protection questions addressed to it "completely unsatisfactorily". For example, the companies had claimed "that they do not operate in Italy and that European legislation does not apply to them". The supervisory authority, which initially also took action against OpenAI's ChatGPT, also launched an official investigation. German data protection experts also want to scrutinize DeepSeek. They complain that the operator allows itself extensive access to IP addresses, chat histories, uploaded files and even the pattern and rhythm of keystrokes.

In its privacy policy, DeepSeek does not conceal the fact that the company stores user data on Chinese servers and processes it in accordance with national law, which stipulates cooperation with the country's intelligence services. However, this has not stopped employees of the US Department of Defense from trying out the service from their work computers for at least two days, reports the financial service Bloomberg. The Pentagon is said to have since begun blocking DeepSeek in parts of its network.

According to CNBC, the US Navy already prohibited its employees from accessing R1 on January 24 due to security and ethical concerns. The US government as a whole is still examining the potential impact of the growing interest in the Chinese service on national security.

(nen)