Because you're talking about an extreme edge case.
The news articles I read stated that the carbon monoxide poisoning was a theory postulated by a member of the family i.e. bereaved, with no other source attributed. That person is not indicated to have any medical or technical qualifications to make such an assessment.
That said, it's not a terrible theory IMO. The snow got pretty high, and it's likely that it covered up the car's normal air inputs. However there was still void space under the car, particularly as the exhaust pipe melted snow there. The engine would tend to draw air from that under-space as its normal inputs got covered by snow. Likewise the exhaust could still discharge into this under-the-car space.
Especially if they plowed into a snowbank (covering it on front and sides) and then got a couple inches more snow, the only route for engine air may have been from the rear down that under-car tunnel.
So you'd have an edge condition where the car is operating in a confined space with poor air circulation; CO2 isn't being carried away and fresh oxygen isn't being replaced. Exhaust is very close to 0% oxygen if things are working properly, all replaced with CO2. While other air is being mixed in, the air is becoming richer and richer in CO2 and depleted in oxygen.
Cars are designed to be run outdoors where we can count on ~21% oxygen and a tiny fraction of 1% of CO2. The system is built to assume normal atmosphere blend; cars do not have HC, CO and NoX sensors in the exhaust stream; hence the semiannual ritual of smog checks :) Their only sensor is an oxygen sensor that tells them more or less digitally, "lean" or "not". So cars rely utterly on the assumption that atmospheric mix is near normal. The car would have no way of knowing the atmosphere was corrupted and thus, its mix ratios were incorrect.