Perhaps fortuitous timing for OpenAI that they can spin the fact that defenders have to resort to open Chinese models because OpenAI and Anthropic actively sabotage them with nerfed models into a nice message of making Huggingface part of the privileged group entitled to secure systems.