Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Black Mirror illustrating The Hacked ADI FleetScene still / Black Mirror

The Hacked ADI Fleet

Autonomous Drone Insects built for a useful public purpose are remotely compromised and repurposed as weapons against people selected at national scale.

A useful autonomous fleet is remotely compromised and repurposed at national scale.

This is an external security compromise of a deployed system, not misalignment arising from the fleet’s learned objective.

AI Risk families

This scenario is an example of the following types of AI risk.
  1. Malicious use
  2. Security
  3. Systemic

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Autonomous Weapons
  2. Algorithmic Monoculture
  3. Cyberattacks