On November 16, the National Security Bureau (NSB) of Taiwan cautioned citizens about using Chinese generative AI language models, stressing the risks of security breaches and misinformation. Recent inspections of five apps—Deepseek, Doubao, Yiyan, Tongyi, and Yuanbao—revealed significant violations across various security indicators, including improper data collection and excessive permissions. Conducted with the Ministry of Justice Investigation Bureau and Criminal Investigation Bureau, these inspections assessed personal data security and generative content risks. Notably, Tongyi failed 11 of 15 security checks, while Doubao and Yuanbao failed 10. The NSB found that these apps often shared biased information reflecting China’s official stance on Taiwan, such as labeling it “an inalienable part of China.” The bureau warned against downloading these apps to safeguard personal data and corporate secrets. Taiwan has already banned Deepseek on government devices since February 2025, but other apps remain unregulated for private use.
Source link
