What label shift means

A spam model was trained when 10% of emails were spam. Now 40% are spam, but spam emails themselves look the same as before. What kind of shift is this?