The main idea is that If we are in a simulation, the most probable type of it either a “Fermi simulation, that is, a simulation which is created by aliens to solve Fermi paradox via simulating possible global risks, or in a “Singularity simulation”,  – a simulation where future AI models its own origin. It means that our simulation will be turned off soon, as at least one of three conditions will be reached:

- It will model a global catastrophe, which is subjectively equal to
the termination.
- it will reach unknown to us goal (but likely related to our AI
ethics) and finish.
- it will reach a nested simulation level, which drastically increases
computational demand and thus the simulation must be turned off.

A.Turchin, M.Batin, D.Denkenberger, R.Yampolskiy. “Simulation Typology
and Termination Risks”.