> wondered whether cancers in humans are a failsafe

I don't think so, but in the context of theories that are more-interesting than "stuff just breaks sometimes", I'd like to suggest the atavistic aspect.

Some cancers resemble a return towards unicellularity, as if "newer" multicellular rules have broken down. Not in any remotely-effective "our genes shall survive the dark times as goo" way, though.