Yeah, I've been hearing that one a lot. I'm always a little suspicious of science journalism (it's usually exaggerated compared to the actual study), so I looked into it.
It turns out the study, performed by Dr. Claudia Kawas, hasn't actually been released yet. But she announced her findings at a recent conference (which is where many journalists have gotten their info). The design was a survey of people aged 90-99 regarding their lifestyle and habits. So, right there, it isn't a very precise study since it's survey-based (ie, reliant upon people giving correct information about themselves). Secondly, the study also found that regular exercise, social and cognitive engagement, and a few extra pounds in older age are associated with longevity.
And anyway, correlation does not equal causation. Many studies have pointed to the fact that socializing is associated with a longer lifespan. I think many who drink in moderation tend to also be more social, which is what confuses the correlation here. I could be wrong, of course.
Either way, even if alcohol legitimately did increase lifespan, the study refers to moderate use, not the excessive and dangerous amount many former drinkers here engaged in.