Federal intelligence and security agencies issued a stark warning Tuesday about what they describe as a systematic campaign by Chinese artificial intelligence companies to extract and replicate advanced American AI technology, raising fresh concerns about intellectual property theft as the United States and China compete for technological supremacy.

The FBI, National Security Agency, and Cybersecurity and Infrastructure Security Agency jointly published findings documenting how six leading Chinese AI developers have mined cutting-edge models from American companies since 2024. The targeted systems include some of the most sophisticated AI platforms available, developed by Anthropic, OpenAI, Google, and xAI.

The practice in question, known as distillation in industry parlance, involves using an advanced AI system to train a new model, effectively transferring knowledge and capabilities without the original developer’s permission. While distillation itself represents a common technique within AI development, its use typically violates the terms of service established by the companies whose models are being exploited.

The federal report identified DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI as the principal Chinese firms engaged in these activities. Notably, the agencies assessed that this campaign has proceeded “likely with Chinese government awareness,” though they stopped short of alleging direct involvement by Chinese intelligence services.

The documented sophistication proves particularly striking. The report details how specific Chinese models distilled particular capabilities from American systems, including specialized functions for legal applications, autonomous agent operations, and interactive assistant features. According to the agencies, the scale and refinement of these efforts suggest that distillation forms not merely a supplementary tool but rather the fundamental core of these companies’ development strategies.

This revelation arrives at a sensitive moment in United States-China relations. President Trump is scheduled to meet with Chinese President Xi Jinping on September 24, and officials from both governments have exchanged sharp criticisms in recent weeks regarding AI technology transfers. The White House has consistently framed artificial intelligence development as a critical race between the two powers, with implications for national security, economic competitiveness, and global influence.

The pattern of allegations extends back at least to last year, when OpenAI first accused DeepSeek of inappropriately basing its model on ChatGPT technology. By February, Google reported facing numerous attempts to clone its Gemini system, though the company declined at the time to identify suspected perpetrators.

The federal report offers American companies specific recommendations for protecting their intellectual property. These include implementing more rigorous verification procedures for users purchasing subscription access to advanced models, sharing information about suspicious behavior patterns across companies, and in certain circumstances, deliberately degrading or altering responses to queries believed malicious in nature.

The White House has elevated this issue to the highest levels of government concern. In April, the director of the White House Office of Science and Technology Policy, Michael Kratsios, circulated guidance to federal agency heads identifying Chinese distillation of American AI systems as a critical threat. His assessment emphasized that these practices allow Chinese firms to benefit from American innovation without bearing development costs while simultaneously circumventing safety measures built into American systems.

The broader accusation fits within a longstanding pattern. The United States has for years accused China of systematic intellectual property theft from American companies across multiple industries. What distinguishes the current situation is the strategic importance of artificial intelligence technology and the potential national security implications of allowing adversaries to rapidly replicate American advances in this transformative field.

China’s embassy in Washington has not yet responded to requests for comment on the federal report’s findings.

Related: Iran Threatens South Korea Over Potential Strait of Hormuz Deployment