Your added paragraph appears to suggest the opposite, that this was an LLM response. Was the "leaked data" a response from an LLM directly?

Yes apparently which makes this report pretty flimsy.

Upthread, OpenAI's security team confirms it's a false report; it's a variant of the empty-prompt hallucination.

Incredible that so many people still don't understand what an LLM is. Especially ones that you would expect to grasp it.