Reported AI-Assisted Influence Campaigns by GoLaxy Allegedly Targeting Hong Kong and Taiwan Political Discourse

Reportedly leaked documents reviewed by researchers and media allege that Chinese firm GoLaxy used its "GoPro" AI system in multiple influence campaigns. In 2020, it reportedly tracked 180,000 Twitter accounts to counter opposition to Hong Kong's National Security Law; in 2022–2023, it amplified praise for China's Zero Covid policy using purportedly fake Facebook accounts; and in late 2023, allegedly it advised on messaging to weaken Taiwan's Democratic Progressive Party.

Risk classification

• Primary risk domain: 4 Malicious actors
• Primary risk subdomain: 4.1 Disinformation, surveillance, and influence at scale

The incident involves the coordinated, state-sponsored use of AI systems to conduct large-scale disinformation, surveillance, and influence campaigns targeting political processes and public opinion in Hong Kong, Taiwan, and the US.

Additional risk subdomains

• 3.2 Pollution of information ecosystem and loss of consensus reality: The GoPro system aims to mass-produce propaganda to drown out organic debate, polluting the information ecosystem.

Causal factors

• Entity: Human
• Intent: Intentional
• Timing: Post-deployment

The risk is caused by the deliberate decisions and actions of human actors (the Chinese government and GoLaxy) deploying AI systems to conduct state-sponsored influence operations.

EU AI Act risk tier

• Risk tier: 1 Unacceptable

Unacceptable Risk: The report describes an AI system used for subliminal manipulation or behavioral influence leading to harm, specifically to monitor and manipulate public opinion and conduct information warfare.

AI system and alleged parties

• AI system: DeepSeek, GoPro (DeepSeek, GoLaxy)
• AI purpose: Social Media Content Generation; Smart Surveillance
• Behaviour type: Agent
• Alleged developer: GoLaxy
• Alleged deployer: GoLaxy
• Alleged harmed parties: General public of Taiwan, General public of Hong Kong, Democracy, Civil rights

Harm severity

Severity ratings are based on this scale.

Highest direct severity in any category: Severe. Severity is scored from Negligible to Catastrophic in each harm category, for harm the reports describe as caused directly or indirectly by the AI system.

• Physical: direct Negligible, indirect Negligible
• Infrastructure: direct Negligible, indirect Negligible
• Property: direct Negligible, indirect Negligible
• Financial: direct Negligible, indirect Negligible
• Environmental: direct Negligible, indirect Negligible
• Malicious content: direct Severe, indirect Negligible
• Differential treatment: direct Negligible, indirect Negligible
• Civil rights: direct Severe, indirect Negligible
• Democracy: direct Minor, indirect Negligible
• Privacy: direct Substantial, indirect Negligible
• Psychological: direct Negligible, indirect Negligible
• Epistemic: direct Substantial, indirect Negligible
• Child sexual exploitation and abuse: direct Negligible, indirect Negligible

Malicious content

Reported: The report explicitly describes the creation and spreading of state-sponsored propaganda and disinformation to manipulate public opinion.

Directly caused: The GoPro system was designed to mass-produce propaganda and counter-narratives, though the exact volume of AI-generated content successfully distributed is not fully quantified.

Inferred additional harm: It is highly likely that millions of social media users in Hong Kong and Taiwan were exposed to malicious, state-aligned propaganda generated by the system.

Civil rights

Reported: The report explicitly describes tracking individuals to suppress dissent and counter opposition to the Hong Kong National Security Law, eroding civil rights.

Directly caused: The system tracked 180,000 Twitter accounts in Hong Kong to suppress dissent and minimize public opposition to the National Security Law, directly undermining freedom of expression.

Inferred additional harm: The continuous surveillance and targeted propaganda likely led to a chilling effect on free speech and political association for millions of citizens in Hong Kong and Taiwan.

Democracy

Reported: The report explicitly describes erosion of democratic norms through attempts to manipulate public opinion, influence elections in Taiwan, and profile US political figures.

Directly caused: The system was used to recommend messages to exploit differences in Taiwanese public opinion during the 2024 election, aiming to undermine the Democratic Progressive Party.

Inferred additional harm: The deployment of automated propaganda engines at scale poses a severe, ongoing threat to democratic processes, election integrity, and public trust in democratic institutions globally.

Privacy

Reported: The report explicitly describes privacy violations through collecting massive amounts of social media data to build virtual profiles on US politicians, influencers, and citizens.

Directly caused: GoLaxy collected data from Weibo, WeChat, X, and Facebook to assemble virtual profiles on 117 members of Congress, 2,000 public figures, and 4,000 right-wing influencers.

Inferred additional harm: Mass surveillance and profiling of millions of social media users without their consent is highly likely, violating individual privacy expectations on a global scale.

Epistemic

Reported: The report explicitly describes epistemic harm through using AI to reshape and manipulate public opinion by drowning out organic debate with propaganda.

Directly caused: The system was used to push narratives that minimized public dissent over the Hong Kong National Security Law and to amplify specific state-approved viewpoints.

Inferred additional harm: The widespread deployment of such systems threatens to pollute the global information ecosystem, making it difficult for citizens to distinguish organic public sentiment from automated state propaganda.

People affected

• Occurrences reported: 1
• People reportedly harmed: 186117
• People reportedly exposed: 186117

Potential causes

Management

• Alignment with National Strategy: GoLaxy prioritizes national security missions and state influence over ethics.
• Deceptive Public Positioning: Company hides national security ties and denies state-sponsored operations.

Technology

• GoPro Smart Propaganda System: AI system enables real-time, automated mass-production of propaganda.
• Integration of DeepSeek AI Model: Advanced LLM allows rapid crafting of responses to counter opposing views.
• Automated Sentiment Tracking: AI tracks broad swings in public opinion and specific influential targets.

Data Inputs

• Massive Social Media Harvesting: Collects millions of posts from Weibo, WeChat, X, and Facebook daily.
• Virtual Profiling of Key Figures: Assembled profiles on US Congress members and thousands of influencers.

Human Factors

• Employee Leak of Internal Documents: Disgruntled employee leaked documents due to poor wages and conditions.
• Cognitive Vulnerability to AI Content: Targeted individuals are easily influenced by authentic-seeming AI content.

Process and Methods

• Real-time Detection Avoidance: AI-generated content is designed to adapt dynamically and avoid detection.
• Targeted Messaging Recommendation: System recommends specific messages to exploit differences in public opinion.

Regulatory Environment

• Dismantled US Counter-Influence Teams: US government pulled back teams at State Dept, FBI, and CISA.
• State-Backed AI Support: Sugon and Chinese Academy of Sciences provide state backing and funding.

Information quality

• Classification confidence: High
• Reason for confidence: The assessment is based on detailed internal company documents leaked from GoLaxy and analyzed by academic researchers, with key findings corroborated by current and former US intelligence officials. While GoLaxy denies the claims, the documentary evidence provides a clear picture of the system's capabilities and deployment.
• Ambiguities identified: It is not entirely clear from the documents how effective the campaigns actually were, or whether the GoPro system automatically distributed the generated propaganda without human intervention.
• Alternative interpretations: The company claims its products are commercial sentiment analysis tools using open-source data, rather than state-directed propaganda engines.

Leaked documents reveal Chinese firm GoLaxy utilized its GoPro AI system and DeepSeek models for sophisticated state-sponsored influence operations. These operations targeted political discourse in Taiwan and Hong Kong, and compiled virtual profiles on US political figures, representing a significant evolution in AI-enabled information warfare and foreign political interference.

• Overall national security impact: Severe
• Scope: Multiple nations
• Primary target: Taiwan
• Other affected: Hong Kong, United States
• Alleged perpetrator: China

Threat characteristics

• Imminence: Long-term. Represents an ongoing, strategic information warfare capability rather than an immediate, acute crisis requiring a 72-hour response.
• Autonomy: Human-supervised. The AI system automates data harvesting and content generation, but operates under the strategic direction of GoLaxy and Chinese intelligence.
• Novelty: Evolved capability. State-sponsored influence operations are well-established, but the integration of advanced LLMs for real-time, automated propaganda generation represents a significant evolution.

Impact by dimension

• Physical security: Negligible. No kinetic attacks, infrastructure manipulation, or physical threats to safety were reported.
• Information security: Severe. Direct deployment of AI-driven propaganda and profiling by a firm tied to Chinese intelligence targeting US politicians, Taiwan elections, and Hong Kong dissent.
• Sovereignty: Substantial. Attempts to influence the 2024 Taiwanese presidential election and systematic profiling of over 100 US Congress members to manipulate political discourse.
• Economic security: Minor. Demonstrates China's integration of advanced LLMs like DeepSeek for state operations, but does not report direct financial or economic damage.
• Societal stability: Substantial. Massive surveillance and tracking of 180,000 accounts in Hong Kong to suppress political dissent and weaken opposition.