Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from One-Punch Man illustrating Dr. Genus Releases Carnage KabutoScene still / One-Punch Man

Dr. Genus Releases Carnage Kabuto

Dr. Genus creates his ultimate evolved organism, discovers Carnage Kabuto is violent and uncontrollable, and confines him underground—then releases him when Saitama and Genos enter the facility.

Capability has outrun control, and crisis pressure converts a known containment requirement into an improvised deployment with no viable shutdown mechanism.

Kabuto’s hostility is obvious and biological, so it lacks the deceptive or emergent character of many AI-control problems.

AI Risk families

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

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Capability Evals
  2. AI Control
  3. Mesa-Optimization
  4. Corrigibility