The celebrated Johnson-Lindenstrauss lemma is a powerful tool for dimension reduction via simple (often random) projections that approximately preserve the geometry of the larger dimensional objects. I will discuss an extension of this result to low CP-rank tensors. I show how modewise tensor projections preserve tensor geometry in the analogous way, without doing any initial tensor matricization or vectorization. Time permitting, I will also talk about an application to the least squares fitting CP model for tensors. Based on our joint work with Mark Iwen, Deanna Needell, and Ali Zare.