5 ms·

I remember a blogpost from OpenAI maybe a year ago that went into great detail about how they find these bad actors.