Market will be resolved based on whichever country does one of these things first.
While the punishments for crimes committed by AIs might not be the same as for humans, if AIs are found guilty of crimes in a court of law, or are classified as an entity that can be found guilty of crimes under the law, that will be sufficient.
(While Saudi Arabia has granted citizenship to an AI, Saudi Arabia is an absolute monarchy and citizens do not have voting rights in national elections.)
Related questions
The problems with issuing voting rights to AIs are there might be very many of them, and not all of them might be intelligent enough to sensibly vote.
Moreover, their population growth rate would not be subject to the same fertility and death factors that the human population growth rate in a country is subject to. And in they were treated as cargo for the purposes of moving between countries, then their "immigration rate" could be uncontrolled too.
Making AIs criminally liable for crimes seems much less problematic - they don't necessarily need to be actually imprisoned, as other techniques (RHLF) can be used on them, or they could be simply shut down (which I guess would be similar to the death penalty).
@RobinGreen criminal liability only means anything applied to a human. Ai can vote, you can argue Bush stealing the 2000 election was ai. But you cannot put an ai in a human prison and ai regulations have to happen outside the human criminal system.
@RobinGreen Also, a rich person could "buy votes" by creating sufficiently intelligent AIs and giving them a goal of voting for X (and persuading other voters to vote for X as well). This seems undesirable.