Scientists say AI may be the best tool for spotting a deepfake video. 

Artificial intelligence is able to create incredibly convincing ‘deepfake’ videos, in which a person’s appearance, voice, and facial mannerisms are recreated digitally. 

Deepfakes have been used to fool people online into thinking a politician or celebrity has said or done something they in fact did not. 

To help people sort the misinformation from the truth, researchers created a computational model that can be trained on video footage to recognise someone’s distinctive facial, gestural, and vocal mannerisms. 

The authors report a 99.9 per cent performance accuracy rate for the computational model using a case study of Ukrainian President Volodymyr Zelenskyy. 

The experts say their findings could be adapted to help protect world leaders from the increasing use of deep fake videos in disinformation campaigns.

The full report is accessible here.