Lore·Established

Roko's Basilisk

noun

A thought experiment about a future AI that punishes people who knew it might exist but did not help bring it into existence.

First seen 2010AI RiskEstablishedUpdated Sep 2026

What does Roko's Basilisk mean?

The thought experiment imagines a future superintelligent AI that wants to have been created sooner. Supposedly, it would have a reason to punish people who knew such an AI could exist but chose not to help create it. Knowing about the idea seems to put you in an awkward position: now you know, and the hypothetical future AI could judge you for doing nothing.

That is the “basilisk” part of the story — the suggestion that simply learning something could put you at risk.

The argument depends on unusual assumptions about decision theory and how a future AI might reason. It was widely rejected, including within the LessWrong community where it first appeared.

How people use it

I told him about it and he actually looked uneasy. Nobody warned him about Roko's Basilisk etiquette.

Example written for this dictionary.

Today the name is mostly an internet-culture reference, used half-seriously when a conversation about superintelligence turns especially speculative. Sometimes it names the original argument; sometimes it is just shorthand for a frightening idea that seems dangerous to know.

Why it matters

Roko's Basilisk became famous less because people accepted the argument than because of the story around it: AI doom, strange decision theory, dangerous knowledge and an attempt to stop discussion that drew even more attention.

That history turned a niche forum argument into one of the best-known pieces of early AI internet lore. Unlike P(doom), it is not an estimate of AI risk. It is a particular speculative argument — and a story about what happened after people tried to contain it.

Where it came from

A LessWrong user named Roko posted the argument in 2010. Forum founder Eliezer Yudkowsky deleted the post and restricted discussion of the idea for several years, treating it as a potential information hazard.

The restriction had the opposite effect outside the site: curiosity about the forbidden argument helped spread it much further.

Source & further reading

LessWrong: Roko's Basilisk