Last updated:
A regulatory remedy requiring a company to delete AI models or algorithms that were developed using improperly or unlawfully acquired data.
Pioneered by the US FTC, this death penalty for algorithms goes beyond simple fines. It forces organizations to destroy the weights and parameters of models if the underlying training data violated privacy or consumer protection laws. This makes data provenance an existential risk for AI development, as years of R&D can be erased by a single enforcement action.
Real world example:
In the Rite Aid settlement, the FTC required the pharmacy chain to delete any facial recognition models derived from customer images that were collected without adequate notice or bias testing.
