OpenAI, Google, and Anthropic Unite to Combat AI Model Copying by Chinese Competitors

Silicon Valley Rivals Unite Against AI Model Copying

In a striking shift from fierce competition to collaboration, OpenAI, Google, and Anthropic have begun working together to combat AI model copying by Chinese AI competitors. The three companies are sharing information through the Frontier Model Forum, an industry nonprofit they founded with Microsoft in 2023, to detect adversarial distillation attempts that violate their terms of service1

1

. This OpenAI, Google, Anthropic collaboration marks a rare instance where Silicon Valley rivals unite to address what they describe as both an economic threat and a national security risk3

3

.

Understanding Model Distillation and the Economic Impact

Model distillation is a technique where a smaller, more efficient "student" model is trained to mimic the behavior of a larger "teacher" model, often at a fraction of the original development cost. While some forms of distillation are widely accepted—such as when companies create smaller versions of their own models—the practice becomes controversial when used by third parties to replicate proprietary work without authorization1

1

. The unauthorized copying of AI models involves repeatedly querying powerful systems, collecting answers, and recreating similar behavior—a shortcut that experts say erodes years of research and development4

4

. US officials estimate that this practice costs Silicon Valley labs billions of dollars in annual profit, creating significant economic pressure on American AI innovation1

1

.

DeepSeek Allegations Spark Intensified Scrutiny

The issue of adversarial distillation drew significant attention in January 2025 following DeepSeek's surprise release of the R1 reasoning model. Microsoft and OpenAI subsequently investigated whether the Chinese startup had improperly extracted large amounts of data from US models to create R11

1

. In February, OpenAI warned US lawmakers that DeepSeek had continued using increasingly sophisticated tactics to extract results from American models, claiming the Chinese firm was relying on distillation to develop new versions of its chatbot. OpenAI accused DeepSeek of attempting to "free-ride on the capabilities developed by OpenAI and other US frontier labs" in a memo to the House Select Committee on China1

1

.

National Security Risk and Price Undercutting Concerns

Beyond economic losses, US AI companies warn that imitation AI models pose serious national security risks. Leading labs express concern that foreign adversaries could use distillation techniques to develop AI models stripped of safety guardrails, such as limits preventing users from creating deadly pathogens1

1

. Most models made by Chinese AI labs are open weight, meaning parts of the underlying AI system are publicly available for free download and use, enabling significant price undercutting1

1

. This poses an economic challenge for US companies that have kept their large language models (LLMs) proprietary, betting customers will pay for access to help offset the hundreds of billions spent on data centers and infrastructure1

1

.

Information Sharing Strategy Mirrors Cybersecurity Industry

The information sharing approach by US AI companies echoes standard practices in the cybersecurity industry, where firms regularly exchange data on attacks and adversaries' tactics to strengthen network defenses. By working together, the AI firms seek to more effectively detect the teacher-student model exploitation, identify who's responsible, and prevent unauthorized users from succeeding1

1

. The Trump administration has signaled openness to fostering such cooperation, with President Donald Trump's AI Action Plan calling for creation of an information sharing and analysis center partly for this purpose1

1

.

Challenges and Future Implications for the Global Tech Race

Despite the collaborative effort, information sharing remains limited due to AI companies' uncertainty about what can be shared under existing antitrust guidance while countering the competitive threat from China1

1

. This tension between intellectual property protection and regulatory constraints will likely shape the global tech race moving forward. The allegations that Chinese companies breach terms of service by generating massive outputs to train their models highlight growing geopolitical tension in AI development2

2

. As the AI race intensifies, observers should watch for potential policy changes that could enable more robust cooperation among American firms while navigating competitive concerns in Silicon Valley.