Skip to content
Submit incident
AII-1002-08093
ReportedReported by fewer than three independent publishers, with no official finding.

Chinese AI company targeted 2024 Taiwan election and Hong Kong dissent with bot networks

According to documents discovered by Vanderbilt University researchers, GoLaxy, a Chinese AI company founded by the Chinese Academy of Sciences, used generative AI and psychological profiling to target Hong Kong and Taiwan with bot networks and personalized propaganda. The company identified dissidents in Hong Kong and suggested methods to weaken Taiwan's Democratic Progressive Party ahead of the 2024 election, while accumulating extensive data on both territories and US political figures.

5 August 2025 · 中科天璣 (GoLaxy) · Misinformation and synthetic media
5 August 2025
AII-1002-080932 sources ↓
Quick facts
Category
Misinformation and synthetic media
Country
Taiwan
Source
New York Times (Chinese)
Date
5 August 2025
Drafted by
Raidu Incident Detection

What happened

According to documents discovered by Vanderbilt University's National Security Institute, GoLaxy, a Chinese company founded in 2010 by a Chinese Academy of Sciences research institute, deployed generative AI and psychological profiling technology against Hong Kong and Taiwan.

The company identified thousands of protest participants and opinion leaders from 180,000 Hong Kong Twitter accounts and used bot networks to "correct" messages it deemed lies and misinformation in order to suppress opposition to the 2020 national security law. Ahead of the 2024 Taiwan election, GoLaxy suggested methods to weaken the Democratic Progressive Party and recommended deploying bot account networks to exploit partisan political divisions.

Reported impact

GoLaxy accumulated extensive data on Taiwan including government organizational charts with political leanings, attitudes toward China and GPS coordinates, as well as personal profiles of over 5,000 Taiwan accounts. Documents suggest the company also collected data on at least 117 US Congress members and over 2,000 US political figures and thought leaders, indicating possible plans for expansion to the United States.

GoLaxy's AI-generated messages are designed to appear realistic and are delivered at unprecedented scale, gradually infiltrating daily digital discussions in ways users may not detect. The technology combines generative AI with mass individual data to build dynamic psychological profiles and generate customized synthetic content in real time.

Index analysis

GoLaxy is connected to China's political system through its founding by a Chinese Academy of Sciences institute, ongoing collaboration with high-level intelligence and party-military departments, and 2021 funding from Sugon, a supercomputing company listed by the Pentagon as affiliated with Chinese military. The company's strategic positioning connects it directly to generative AI systems and China's leading AI models.

The technology weaponizes the openness, transparency and pluralistic debate that underpin democratic societies. As the boundary between surveillance and persuasion erodes, AI-generated content can propagate silently at massive scale without significant resistance, continuously shaping public opinion and corroding democratic institutions in ways larger in scope and more sophisticated than current social media platforms.

Governance controls

GoLaxy denied in a written statement that it provided technological support for activities in Hong Kong and Taiwan, denied developing personalized bot account networks or psychological profiling tools, and denied being affiliated with any government institution or organization.

A GoLaxy representative stated the company's business focus is providing commercial intelligence services and denied collecting data specifically about US officials. The documents discovered did not include specific examples of individual conversations generated by the technology.

Classification

Category
Misinformation and synthetic media
Industry
Media, social platforms and marketing
Country
Taiwan
Harm status
Actual harm
Intent
Intentional
Timing
Unknown
Harm types
Psychological
Harmed
General public, Civil society
Failure modes
Manipulation or dark pattern, Fabricated citation
AI systems
GoLaxy AI platform, DeepSeek-R1

Sources and evidence

This record was researched and written by the Index, based on the following original reporting.

New York Times (Chinese)
中國的新「武器」:用人工智慧發起宣傳戰
6 August 2025
Liberty Times Net
中國AI認知作戰黑手伸入台灣!美研究揭中企2024大選前已發動攻擊 - 國際 - 自由時報電子報
6 August 2025