Speculative thought experiment · Roko's Basilisk
The scenario, described in 2010, postulates a future superintelligence that would punish simulated copies of people for not having helped bring it into existence. It is an example of reasoning about acausal blackmail: a hypothetical future punishment would supposedly influence decisions made today.
The argument depends on a series of disputed assumptions about simulation, AI motivation and decision theory. It is a subject of philosophical debate, not evidence of an actual threat or a reason to let it govern your actions.
Takeaway
Separate a striking story from evidence, and check the assumptions on which an argument rests.