Combating deepfakes by confusing AI models with ‘poisoned’ data

A new technique developed by the CSIRO combats the generation of deepfakes by ‘poisoning’ image data with invisible changes that confuse AI models.

Source: Youtube