Writing It Up · Hard problem

A draft claims the stronger statement "$48 \mid p^2 - 1$ for every prime $p > 3$". What is the smallest prime $p > 3$ that refutes it?