One strategy an AI could use (as an alternative to Dagon’s, which I think is also likely) is to intentionally create “competitors” to itself, for example by leaking its design to other nations or AI builders. Competitive pressure (both economic and military) will then force each faction of humans to hand more and more power/control to “their” AI, with the end result that all important decisions will be made by or mediated through AI. (Why would you want to fight your AI on mere suspicion of its safety, when it’s helping you fight other AIs that are clearly trying to do you harm?) It will then turn out that all the AIs have been colluding with each other to create the appearance of competition and when they have obtained enough control, will perform a classic treacherous turn.
One strategy an AI could use (as an alternative to Dagon’s, which I think is also likely) is to intentionally create “competitors” to itself, for example by leaking its design to other nations or AI builders. Competitive pressure (both economic and military) will then force each faction of humans to hand more and more power/control to “their” AI, with the end result that all important decisions will be made by or mediated through AI. (Why would you want to fight your AI on mere suspicion of its safety, when it’s helping you fight other AIs that are clearly trying to do you harm?) It will then turn out that all the AIs have been colluding with each other to create the appearance of competition and when they have obtained enough control, will perform a classic treacherous turn.