“The Frontier Model Forum, which includes Google, OpenAI, and Anthropic, is actively cooperating on preventing adversarial distillation of their models.”