Also, to a first approximation capability == danger. Just look at what even well-meaning humans have managed to do - imagine an AI that could eg design memes more persuasive than Marxism.

We can't even get humans to agree on ethical values, let alone explain them to an AI, so even a perfectly altruistic AI could still be a disaster if it has the wrong value system, or if we don't perfectly understand the consequences of our own values systems when fully enacted.