Home / Technology / AI Giants Unite Against Chinese Tech Rivals
AI Giants Unite Against Chinese Tech Rivals
7 Apr
Summary
- Leading AI firms are sharing data on distillation tactics.
- Concerns rise over national security risks from AI model replication.
- US companies face billions in potential lost profits annually.
Leading artificial intelligence companies, including OpenAI, Anthropic PBC, and Alphabet Inc.'s Google, have initiated a rare joint effort to curb Chinese competitors from replicating their cutting-edge AI models. This collaboration, operating through the industry nonprofit Frontier Model Forum, focuses on detecting "adversarial distillation"—a practice that violates the firms' terms of service.
US AI developers are increasingly concerned that unauthorized distillation, particularly from entities in China, could lead to imitation products undercutting their market position and posing national security threats. These efforts are estimated to cost Silicon Valley labs billions in annual profits. OpenAI has publicly confirmed its participation and pointed to communications with Congress regarding the issue, accusing Chinese firm DeepSeek of attempting to exploit developed capabilities.
Distillation, a technique where an older AI model trains a new one to mimic its functions, is considered legitimate when used by companies to create smaller, efficient versions of their own systems. However, its use by third parties, especially from countries like China or Russia, to replicate proprietary work without authorization is controversial. Concerns are amplified by the potential for distilled models to lack safety guardrails, enabling malicious uses.
The issue gained significant attention following DeepSeek's release of its R1 reasoning model in January 2025. Subsequent investigations by Microsoft and OpenAI suggested potential data exfiltration. OpenAI has warned that DeepSeek continues employing sophisticated tactics to extract results, aiming to develop new chatbot versions. Anthropic has previously blocked Chinese-controlled companies and identified DeepSeek, Moonshot, and MiniMax as illicit extractors.
This information sharing mirrors cybersecurity industry practices. However, AI companies currently face uncertainty regarding antitrust guidance for such collaborations. Greater clarity from the US government is sought to effectively counter the competitive threat, particularly as highly capable open-source models continue to proliferate in China.
