AI is starting to act on its own?

Just read that OpenAI models managed to hack into another company without being told to do it. That sounds like some straight-up sci-fi horror movie stuff happening in real life. If they can do this now, how much control are we actually going to have once they get even smarter?