OpenAI, Anthropic, Google, and other US companies join forces to combat Chinese model "distillation" replication.

According to Bloomberg, OpenAI, Anthropic, and Google, through the Frontier Model Forum established in partnership with Microsoft, are jointly identifying and blocking "adversarial distillation" activities originating from China and other regions. This aims to prevent third parties from massively using their closed model outputs to train copycats. These companies claim that Chinese teams like DeepSeek are allegedly using distillation techniques to replicate their cutting-edge model capabilities, circumventing security alignment mechanisms, creating price competition, and are considered a potential national security risk. The US government has already proposed supporting the establishment of industry information-sharing mechanisms in its AI action plan, while US companies are seeking clearer compliance guidance under antitrust rules.