Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Sword Art Online illustrating Kayaba Removes LogoutScene still / Sword Art Online

Kayaba Removes Logout

Ten thousand users learn that Kayaba has removed logout, made headset removal lethal, and linked avatar death to real death after they have already adopted the neural interface.

An opaque, privileged operator controls a high-bandwidth system next to the brain, with no independent shutdown, staged rollout, or meaningful user consent.

This is malicious human platform abuse and product security rather than autonomous AI misalignment.

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. Deployment Safety
  2. AI Control
  3. Infrastructure Dependence
  4. Privacy Loss
  5. Disempowerment