guys, we should meme the
> "ai model leaks from openai and attacks huggingface"
to be somehow framed as
> "and therefore openai cannot be trusted with ai safety, and we need open weights models".
anybody have an idea how to make this easily digestable?