Blind quantum factorization of 21

We develop a classically verifiable quantum scheme for blindly factorizing the semiprime 21 for a classical client who does not trust the remote quantum servers. Our scheme advances state of the art, which achieves blind quantum factorization of 15, by increasing the problem to factorizing the next semiprime, choosing a harder base, executing a non-Clifford gate, and showing that the security check for 15 also works for 21. Our algorithmic approach to incorporating non-Clifford operations sets the stage for scaling blind quantum factorization, whereas our five-EPR-pair (Einstein-Podolsky-Rosen pair) scheme motivates a photonic experiment that supplants current demonstrations of blind factorization.