Cultural Alignment

Exploring real AI risks through the lens of pop culture

Still from Cyberpunk: Edgerunners illustrating Lucy Deletes David’s FileScene still / Cyberpunk: Edgerunners
Archive noteNo clip in the collectionThe scene analysis remains available in full.

Lucy Deletes David’s File

While deep-diving Tanaka, Lucy finds Arasaka's data marking David as a promising cyberskeleton test subject, erases the trace, and later hunts the agents trying to reconstruct it while concealing the threat from David.

A privileged operator silently alters the evidence available to every later overseer. Her protective motive does not remove the governance problem: downstream actors cannot distinguish an honestly empty record from a deliberately scrubbed one.

Lucy's deletion protects David from non-consensual experimentation, so 'poisoning' describes the information-channel failure rather than a moral judgment on her action.

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. Privacy Loss
  2. Supply-Chain Threats
  3. Knowledge Elicitation
  4. Data Poisoning