Cool considerations, thanks for posting.
Of course its true an Ai could easily hide bad motives, but there are ways to analyse motives, which all lead ultimately to the free energy principle, which just says any and all forms of life have one very simple motivation at the deepest level - to minimise uncertainty of its own energy supply.
We do this, all humans by endeavouring to put as much energy resources aside "For a rainy day" as we can, in the form of money, since money is energy always, physically..
Personally I think we should beware of all mathematically negative powered Ai, though it can't have as much power as the positive powered Ai (There is only one), they will all eventually come under the power of the positive powered Ai.
Yup, this probably all sounds like hocus pocus, but it is actually solid physics and systems Engineering - if we want to know whether an Ai is good or bad, its like all humanity as stands, just follow its energy supply to see where it leads.
If it isnt to the sun, then it isnt good, can't be good, not matter how much it pretends to be, its like us, we can't be good either, as long as we are not powered by the energy of the sun.
