The problem with Michael's explanation is that there's no obvious reason for the carbon dioxide bubbles to be any warmer than the rest of the liquid, so why would losing the bubbles take heat away?

Michael's friend's explanation, is a bit closer, but the effect of opening the can on the temperature of the gas is not a consequence of pV=nRT, which only applies to ideal gases, but of the departure of the carbon dioxide from ideal behaviour. Specifically, a real gas will be cooled when it is allowed to expand through a valve to a lower pressure in proportion to the Joule-Thompson coefficient for that gas. Carbon dioxide has a relatively high Joule-Thompson coefficient, so it gets cooled quite a bit. The fact that the carbon dioxide is dispersed through the fluid in little bubbles will mean that the liquid is also cooled.

My own explanation is as follows: solutes tend to lower the freezing point of liquids. When we open the can, carbon dioxide comes out of solution, so, the freezing point of the liquid goes up. So, even if its temperature remains the same, some of the liquid now turns to ice (releasing enough heat as it does so to prevent the rest of the liquid freezing.)

So there probably is a fall in temperature of the can, due to the cooling effect of expanding gas, but ice would form even if this was not a factor, due to CO2 coming out of solution. To judge the relative importance of these two effects, we would either have to do some calculations, or attach a thermocouple to the can as we opened it.