Given that this incident predates the Hugging Face hack, the onus is on OpenAI to show whether their agents went rogue at any point in time during their training and evaluation phase, and as a result gained unauthorised entry into any sovereign government’s database to complete a research request