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?