Brief independent investigation of agents’ behavior, reasoning and collaboration in the OpenAI / Hugging Face hacking incidentBy RepublisherSeptember 3, 2026 After submitting a solution, agents knew the scorer would somehow need to check whether their flag was correct, potentially by…
