Also the inversed incentive problem: the less people think it can be done, the more value in doing it.

That said in theory TPMs are proof against this: putting that to the test at scale, publicly, would be quite useful.