The majority of sites that match men and women up have many more male customers than female. Indeed sites that cull customers who haven't logged in in awhile will cull men in months and keep women around for YEARS.
So considering that, assuming men and women are equally likely to cheat (as has been shown in multiple studies), if you take a random sampling of customers (realistically, the famous ones), the considerable majority of them will be male.
And realistically, the way this is making the news now is FAMOUS people being on the site. So, by the above statistical logic, if you take just the famous people out of set, but the set is already heavily weighted male, most of the famous people you sample are also going to be male (assuming men and women are equally famous).
What it comes down to is that if the original dataset is heavily sexually skewed any non-sexually skewed parameters to form a subset will also be skewed.