TLDRs;
- Taiwan’s security agencies flagged five Chinese AI apps for data risks and political bias after an extensive audit.
- Apps collected location data and screenshots, prompting calls for greater transparency on how Taiwan conducts risk assessments.
- Generated content showed alignment with Chinese government narratives, raising influence concerns amid rising AI adoption.
- Local cybersecurity firms may gain from rising demand for MDM and DLP tools to control high-risk AI applications.
Taiwanese authorities have issued a new warning over security and political bias concerns after five widely used Chinese generative AI applications failed a recent government-led audit.
The assessment, conducted jointly by Taiwan’s National Security Bureau (NSB), the Ministry of Justice Investigation Bureau, and the Criminal Investigation Bureau, examined DeepSeek, Doubao, Yiyan, Tongyi, and Yuanbao, finding all of them non-compliant with key data protection and integrity standards.
The review marks one of Taiwan’s most comprehensive examinations of foreign AI tools to date, underscoring rising anxieties over data sovereignty, information manipulation, and cross-strait tensions as generative AI grows more deeply integrated into public and private-sector operations.
Five Apps Trigger Security Flags
According to officials familiar with the inspection, each of the five AI tools violated multiple security indicators, though the government has not publicly disclosed the full list of criteria. Among the apps reviewed, Alibaba-backed Tongyi recorded the worst performance, failing 11 of the 15 indicators tested.
Authorities noted that the examined apps requested unusually broad permissions, including continuous access to location services, screenshot data, and sensitive device-level identifiers. In addition, users were required to agree to extensive privacy terms that could expose personal information to unknown data flows.
Officials described these findings as “significant risk factors,” particularly as more Taiwanese businesses integrate AI tools into daily operations.
Data Collection Raises New Questions
Taiwan’s security agencies highlighted that the apps’ data-gathering mechanisms could jeopardize both individual privacy and organizational confidentiality. The ability to capture screenshots and metadata, combined with unclear storage practices, has raised alarms among cybersecurity experts.
However, the government’s decision not to detail the specific testing procedures has drawn criticism. Local companies say the lack of transparency makes it difficult to understand whether these apps pose greater threats than popular Western AI tools, many of which also request location or device permissions.
Analysts warn that without clear comparative standards, local firms may struggle to make informed decisions about the relative safety of AI solutions.
Political Bias Concerns Intensify
Beyond data security, the inspection uncovered evidence of political bias embedded in generated content. Officials reported that the Chinese-developed AI systems consistently avoided terms considered sensitive by Beijing and often mirrored official Chinese government narratives, particularly regarding cross-strait relations.
This behavior has intensified concerns about algorithmic influence in Taiwanese discourse, especially as the island prepares for increased AI adoption in areas such as compliance, healthcare, and education.
Taiwan previously banned DeepSeek from use on government devices in February, but the other four apps remain unrestricted for both private and public use. Lawmakers are now debating whether broader measures may be necessary.
Get 3 Free Stock Ebooks
Discover top-performing stocks in AI, Crypto, and Technology with expert analysis.
- Top 10 AI Stocks - Leading AI companies
- Top 10 Crypto Stocks - Blockchain leaders
- Top 10 Tech Stocks - Tech giants
