“OpenAI is conducting research on how to maintain safety for powerful AI models after they have been open-sourced, with potential findings to be shared next year.”