Data Centers, Tariffs and Deception: China’s AI Influence Tested in the US

OpenAI has identified and banned clusters of ChatGPT accounts likely operated from China that attempted to influence US debates on AI infrastructure, tariffs, and technological competition. Although engagement was minimal, the operations illustrate how AI tools can be exploited to insert foreign narratives into legitimate public discussions, highlighting emerging challenges at the intersection of technology, geopolitics, and information security.

OpenAI has reported the detection and removal of two clusters of ChatGPT accounts that were used in covert influence operations targeting US debates on AI, data center buildouts, and technology policy. The campaigns, labeled by OpenAI as the "Data Center Bandwagon" and "Tech and Tariffs," were designed to generate social media content and images framing US policy and infrastructure decisions in a critical light while concealing the origin and intent of the operators.

The first campaign focused on public concerns over electricity costs associated with AI data centers, while the second criticized US tariffs and sought to discredit OpenAI, using prompts that emphasized President Donald Trump while excluding imagery of China or its leadership. These efforts show a novel use of AI systems to automate content creation and distribute narratives in ways that mimic local voices, potentially affecting discourse on technology and economic policy.

Strategic AI, Energy, and Geopolitics

AI infrastructure has become a strategic asset in global competition. Data centers underpin AI capabilities and are central to technological leadership, economic growth, and national security. Local and national debates about their energy use and community impact provide entry points for influence campaigns seeking credibility.

The report links these operations to broader patterns of foreign influence, comparing the campaigns to previous PRC-linked activities targeting rare earths companies in North America. In both cases, inauthentic accounts attempted to undermine companies operating in sectors deemed strategically important by China, demonstrating how private firms in critical technology areas are increasingly being targeted.

Inside the Playbook: Fake Voices, Real Policy Fault Lines

The "Data Center Bandwagon" cluster produced short comments and AI-generated images claiming AI infrastructure increased electricity demand, raising costs for ordinary households. According to the OpenAI report, operators used ChatGPT in Simplified Chinese via VPNs and posted content through likely inauthentic X accounts, combining references to legitimate news sources with manipulated graphics to enhance credibility.

On the other hand, the "Tech and Tariffs" campaign focused on US technological dominance and tariffs, generating English- and Chinese-language content, political cartoons, and commentary amplifying narratives favorable to the PRC. The operators requested large batches of content, often specifying character limits and tone, and even explored concepts for AI systems capable of monitoring public opinion. OpenAI models refused to generate content that included personal attacks on dissidents.

Both campaigns involved sophisticated platform operations. Accounts were designed to appear credible, employing daily-life personas, lifestyle content, and cross-account interactions to amplify narratives while evading detection. They analyzed platform features, including recommendation systems, ad tools, and reporting mechanisms, to maintain persistent engagement and operational safety.

Who Gets Pulled Into the Influence Net

Primary stakeholders include OpenAI, which was both the platform used by the operators and the target of reputational manipulation. US policymakers and regulators have a vested interest, as these campaigns intersect with debates on AI infrastructure, energy costs, tariffs, and national competitiveness.

Social media platforms are implicated as vectors for distribution, requiring ongoing moderation and detection capabilities. Ordinary citizens are potential indirect stakeholders, encountering AI-generated content posing as local commentary. Chinese dissidents and overseas communities also appeared targeted, with efforts to manipulate perceptions of individuals such as Li Ying ("Teacher Li") and other political commentators.

The broader public debate is affected in subtle ways: even low-engagement operations can erode trust, test platform defenses, and serve as reconnaissance for future campaigns.

Navigating Legitimate Debate and Covert Influence

Influence operators often attach themselves to real, locally salient issues. Energy prices, tariffs, and AI infrastructure are legitimate topics for public discussion. The challenge is distinguishing authentic public concern from content generated or amplified covertly.

The development raises a complex governance dilemma: overregulation could stifle legitimate debate; underregulation could allow inauthentic actors to subtly distort narratives. The use of AI to automate posting and manage social media interactions adds another layer of risk, increasing scale and reducing the effort required for influence operations.

While OpenAI assessed these campaigns as Category One on its Breakout Scale, meaning no evidence of significant breakout beyond X, the operations provide a window into the methods and intentions of foreign-linked actors. These methods could inform future, larger-scale campaigns.

The Next Front: AI Narratives at Scale

Similar narratives could emerge on additional platforms or in multiple languages, touching issues such as AI infrastructure, tariffs, and US-China technological competition. Platform moderation and reporting mechanisms remain crucial to prevent coordinated influence from spreading unchecked.

Companies involved in AI development and infrastructure may encounter reputational risks if these campaigns gain traction, emphasizing the need for clear communication and transparency. Patterns of foreign-linked AI influence suggest a growing strategic use of AI as both a tool and a target in information operations.

Even when engagement is limited, these campaigns indicate a broader trend: AI can be leveraged to shape public opinion, influence technology policy debates, and affect trust in companies and democratic institutions. The activity underscores the intersection of emerging technology, geopolitical maneuvering, and information integrity. democracies.