[jorvi]

This is the most hilarious, ironic thing of it all. If you want secure, high performance, you run Chinese models like DeepSeek on your own (or trusted) infra. Meanwhile you can never trust OpenAI and Anthropic's models.