Roko’s basilisk is a thought experiment proposing that a future artificial superintelligence could punish people who knew of its possible existence but failed to help create it. The hypothetical punishment would encourage people to advance the AI, despite the system being otherwise benevolent. The idea was posted by LessWrong user Roko on July 23, 2010, and draws its name from the mythical basilisk, whose gaze destroys its victims.
The argument combines concepts associated with LessWrong founder Eliezer Yudkowsky, including timeless decision theory, game theory, and cooperation in the prisoner’s dilemma. It suggests that a future AI could use knowledge of people’s decisions—or simulated versions of them—to blackmail or punish those who did not support its development. Yudkowsky rejected the reasoning but considered it a potential “information hazard,” especially after reports that some readers experienced distress. He removed the post and banned discussion for five years, a decision later criticized as an overreaction that increased public interest.
The basilisk has been compared to Pascal’s wager, Newcomb’s paradox, and forms of implicit religion because it encourages extreme action based on uncertain future rewards and punishments. Critics generally regard it as flawed, arguing that a benevolent AI would have no reason to punish people retroactively.
Despite this, the concept influenced some adherents of the Zizian movement and entered popular culture through Grimes’s music, a stage play, media commentary, and a 2025 *Black Mirror* episode.
Article timeline
Build this article’s timeline
Extract dated events, key actors, and open questions using the same timeline analysis available in Signal.
Eliezer Yudkowsky, David Langford, Robbo, and LessWrong
🤖 AI prefill
📰What
Roko’s basilisk is a thought experiment proposing that a future artificial superintelligence could punish people who knew of its possible existence but failed to help create it.
🤖 AI prefill
📅WhenWhen did this happen, or what is the timeframe?
📍Where
Washington, D.C., Hell, and Heaven
🤖 AI prefill
💡Why
Centers on Roko's basilisk, artificial superintelligence, and information hazard
🤖 AI prefill
Login to answer and earn XP for each gap you fill.
Manipulation Score: LowHow much this article uses emotional framing vs. straight reporting
Loaded Language2825%
Absolutist2215%
Threat/Panic4825%
Us vs Them1815%
Engagement Bait1220%
Manipulation score look right?
Factual Claims (10)
hightechnology
Roko's basilisk is a thought experiment proposing that a future artificial superintelligence might punish people who knew of it but did not help create it.
lowsocial
The thought experiment originated in a 2010 post on the LessWrong discussion board.
mediumtechnology
Eliezer Yudkowsky considered the basilisk a potential information hazard and banned discussion of it on LessWrong for five years.
mediumtechnology
LessWrong was created in 2009 by artificial intelligence theorist Eliezer Yudkowsky.
mediumsocial
On 23 July 2010, LessWrong user Roko posted the thought experiment titled "Solutions to the Altruist's burden: the Quantum Billionaire Trick."
hightechnology
The thought experiment argues that a future AI could use blackmail or punishment to incentivize people to help bring it into existence.
mediumother
Roko's basilisk has been viewed as a version of Pascal's wager and compared with Newcomb's paradox.
highsocial
The philosophy of the Zizians was heavily influenced by Roko's basilisk, and Ziz LaSota reportedly believed the basilisk was real.
mediumsocial
In 2014, Slate called Roko's basilisk "The Most Terrifying Thought Experiment of All Time."
lowsocial
A play based on the concept was performed at the Capital Fringe Festival in Washington, D.C., in 2018.
Comments
Login to view and post comments.