> but I trust them about as much as I trust OpenAI.

So, ideally, not at all?