Two strange ideas that power the scariest thought experiments.
Acausal trade
Can two agents cooperate without ever communicating — even across time or universes — purely by modeling each other? If A can predict that B (who can predict A) will keep a bargain, they can "trade" with no causal contact. It sounds like science fiction, but it follows from certain decision theories where being a predictable agent is rational.
It's the machinery beneath Roko's Basilisk: a future AI "bargaining" with people who can only imagine it.
Infohazards
An infohazard is information that causes harm simply by being known — not misused, just learned. The classic case is Roko's Basilisk itself: the claim that merely understanding the idea puts you at risk.
Most "infohazards" are really just memetic anxiety — scary ideas that spread because they're scary, not because they're true.
Both concepts are more useful as cautionary lenses (on how communities spook themselves, and how predictable agents bargain) than as literal threats.
Related: Roko's Basilisk · Newcomb's Problem and Decision Theory · Pascal's Mugging