Bite Size Comms

Deepfakes are topical with the outcry against deepfake porn images of Taylor Swift circulating on social media. What I would like to discuss is how deepfakes could seriously damage democracy in 2024, a significant year for elections around the world. Sixty-four countries, representing about 49% of the world’s population and including the USA, UK, India, Mexico and Russia, are scheduled to hold national elections this year.

A case study that you may not be aware of. I used to live and work in Slovakia. The country held parliamentary elections in September 2023. It was a very close contest between Robert Fico, a populist pro-Russia politician with a very dodgy track record (he had to resign as prime minister in 2018 over alleged links between the murder of an investigative journalist and his staff) and his rival, Michal Šimečká. Late polls showed Šimečká had a lead. Then, suddenly, an audio deepfake was released two days before the election. The deepfake was widely shared online in Slovakia and seemed to portray Šimečká conniving to rig the vote. The pro-Russian, Fico, won. I met him 20 years ago and was not impressed.

A Microsoft AI model, VALL-E, can clone a voice from just three seconds of recordings. 

In January, Fenimore Harper – a communications agency, found 143 deepfake video adverts of Rishi Sunak on Facebook over one month that made up various crises. These reached 400,000 people and cost £12,929, or look at that as profit to Meta. 

In a very bizarre press conference late in 2023, Vladimir Putin fielded a videocall from a deepfaked copy of himself. “Do you have a lot of doubles?” the software doppelgänger asked. The real Putin calmly replied that only one person could speak with the voice of Putin, Putin himself. But honestly, which one was the real Putin – if any, if you read the UK tabloids’ stories about fake Putins?

How to stop these deepfakes? Use your instinct – if it doesn’t look right, it probably isn’t. Flag it as suspicious on social media. Meta, for example, has said it will be more vigilant about sponsored posts with political content this year. That’s going to be primarily AI invigilated. Give the bots a hand! Don’t share deepfake posts, even as a joke. I love a good stunt and turn to @FamousCampaigns for a good, creative laugh. Deepfakes are not that. They were probably created or funded by hostile states or people with weird views on life.

Technology may catch up and help to immediately spot deepfakes. However, it is not right to blame the technology. As Adobe correctly pointed out recently, a human created the AI-supported Photoshop deepfake of Taylor Swift and a human approved it.

This is human-led disinformation.  

[Image: Unsplash]