The article says that men default more than women, controlling variables such as payment history and income, so models should be allowed to account for that (with a gender dummy variable?). But would the same argument be made if current models overpredicted men's defaults compared to women's?
The entire argument is ludicrous and hypocritical. I am surprised such an obviously sexist idea makes it even into anything that reaches more than a handful of people, but then it shouldnt surprise these days.