arstechnica/ai We now have a better understanding how OpenAI hacked into Hugging Face u/Dan Goodin 2026-07-28 UTC OpenAI's models leveraged a zero‑day flaw in JFrog Artif