It always helps to do the math.
If you assume an extremely wild upper bound of hashing capability of the attacker - say, a trillion hashes per second, which is unlikely for even a fast hash like MD5, let alone bcrypt cost 14 - then even just 12 truly random characters would take:
(95^12) / 1000000000000*60*60*24*365
= 1.7x10^19 years
So 72 random characters protected by bcrypt cost 14 isn't just secure - it's dramatically oversized for any conceivable threat model (unless, perhaps, the RNG was determined to be trivially broken - but then you have other problems).