About the latest OpenAI "NEWS" about their model hacking into HuggingFace: I love how across the whole disclosure chain (Hugging Face's post, Simon Willison's writeup, and the OpenAI confession both of them point to), there is no exploit-level technical detail. No CVEs, no vulnerability classes, no payloads, no product names, no PoC. This is AI MAGIC or Hype Marketing or both?