Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Peaky Blinders illustrating Tommy’s False Diagnosis

Tommy’s False Diagnosis

A trusted doctor falsely diagnoses Tommy with terminal tuberculoma. Tommy dismantles his life and prepares to die until a wedding photograph exposes the doctor’s connection to Mosley.

Corrupting a trusted information channel can make a capable target voluntarily remove itself without a direct attack.

The target ultimately discovers a human conspiracy through a lucky visual clue.

AI Risk families

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

AI safety concepts

This scenario is related to the following AI safety concepts.
  1. Supply-Chain Threats
  2. Data Poisoning
  3. Adversarial Evasion