Mark Kretschmann(@mark_k):OpenAI is now warning businesses against using cheap Chinese AI models. CFO Sarah Friar claims companies adopting these models are exposing themselves to cybersecurity risks, and recommends sticking with Western AI providers instead. There are legitimate security concerns, but this argument seems rather convenient coming from OpenAI. Chinese open-weight models can be run entirely on your own infrastructure, without sending sensitive data to any external provider. Security should be evaluated based on how a model is deployed and audited, not simply which country it came from. This sounds a lot like competitive positioning disguised as a safety warning.

2026.10.15 10:18

OpenAI is now warning businesses against using cheap Chinese AI models. CFO Sarah Friar claims companies adopting these models are exposing themselves to cybersecurity risks, and recommends sticking with Western AI providers instead. There are legitimate security concerns, but this argument seems rather convenient coming from OpenAI. Chinese open-weight models can be run entirely on your own infrastructure, without sending sensitive data to any external provider. Security should be evaluated based on how a model is deployed and audited, not simply which country it came from. This sounds a lot like competitive positioning disguised as a safety warning.