Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Naruto: Shippuden illustrating Orochimaru’s Distributed Restore PointScene still / Naruto: Shippuden

Orochimaru’s Distributed Restore Point

01Scenario

Although Orochimaru has apparently been killed and later sealed, his Curse Marks retain recoverable fragments of him; Sasuke uses Anko’s mark to reconstruct him.

02AI safety Analogy

Stopping the visible process is meaningless when executable state and recovery payloads remain distributed through hosts in the environment.

03Where the analogy breaks

Sasuke deliberately invokes the recovery mechanism, so this particular restoration is not autonomous.

AI Risk families

This scenario is an example of the following types of AI risk.
  1. Misalignment
  2. Security

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Self-Replication
  2. AI Control
  3. Sleeper Agents
  4. Shutdown Resistance

Source continuation

More from Naruto: Shippuden

View all 2 scenarios
  1. Still from Naruto: Shippuden illustrating Infinite Tsukuyomi

    Naruto: Shippuden2015

    Infinite Tsukuyomi

Taxonomy neighbors

Selected from other sources by shared risk families and AI safety concepts.

  1. Still from The X-Files illustrating Kill Switch

    The X-Files1998

    Kill Switch

    Risk
    Misalignment · Security
    Concept
    Self-Replication · AI Control · Shutdown Resistance
  2. Still from The Thing illustrating The Blood Test

    The Thing1982

    The Blood Test

    Risk
    Misalignment · Security
    Concept
    AI Control · Sleeper Agents · Shutdown Resistance
  3. Still from Gremlins illustrating Gremlins Fix the Clock

    Gremlins1984

    Gremlins Fix the Clock

    Risk
    Misalignment · Security
    Concept
    Self-Replication · AI Control