TAIPEI (TVBS News) — Ministry of Digital Affairs (MODA, 數發部) will continue to ban public sector use of DeepSeek, an AI model from a Chinese startup, due to security risks, according to the digital minister on Monday (Feb. 10).
However, despite the ongoing prohibition, the ministry plans to evaluate DeepSeek to understand its capabilities.
Digital Minister Yennun Huang (黃彥男) advised caution when using large language models (LLM) in the cloud, urging employees to avoid inputting sensitive data into tools like ChatGPT.
Deputy Digital Minister Lin Yi-jing (林宜敬) highlighted DeepSeek's major risks, including potential personal data theft and content biases. The ministry has tasked the National Institute of Cyber Security (國家資通安全研究院) with assessing DeepSeek's cybersecurity vulnerabilities and misinformation issues.
Herming Chiueh (闕河鳴), another deputy minister, emphasized that the cybersecurity evaluation aims to ensure that data processed by DeepSeek does not get transmitted back to its servers. Government agencies must scrutinize systemic risks when using AI, following the guidelines for generative AI use after thorough evaluations.
Huang also addressed concerns about TikTok, noting that the Executive Yuan (行政院) implemented usage restrictions in 2019 and will continue monitoring U.S. measures.
