Deepfakes: How worried should you be?
by Eliza Bacon
Junior Producer & Editor Global Digital Futures
Hello and welcome to the Global Digital Futures community, and to the first fortnightly column. With every podcast release our editors will be sharing their perspectives on the topic. Keep reading for insights, news and trends.
This week, Chipo interviewed Craig Ryder, whose PhD research is on the impact of fake news on democracy in Sri Lanka. Chipo and Craig’s conversation raises important questions about the impact of deepfakes, with case studies from Sri Lanka, Gabon, Malaysia and more. Listen to it here.
If this is your first foray into the world of deepfakes, you may need a visual introduction. @deeptomcruise is a great place to start - the account currently has one million followers on TikTok and it’s terrifyingly accurate. For an example of how they are already being used in political campaigning, last year the BJP IT Cell produced deepfake videos of Manoj Tiwari to reach different linguistic voter bases.
“If disinformation campaigns are working well enough with poorly doctored images or re-circulated ones, why bother with deepfakes?”
Deepfakes have been around since 2017, when a Reddit user with the username “deepfakes” uploaded a series of videos with female actors’ faces imposed onto pornographic videos. Since then they have become more accurate and easier to produce: there are now apps that make it possible for anyone to create a deepfake and creators are increasingly able to mimic facial, audio and behavioural likeness from just a few hundred images. This means that any of us with enough photos of ourselves on social media are at risk. So, how worried should we be?
Well, it depends. If you are a woman in the public eye it’s not looking great. Non-consensual pornography was the catalyst for the development of deepfakes and it still makes up 96% of deepfakes today. Read about Rana Ayyub’s horrifying experience of being made into viral porn here. Misogynistic harassment and intimidation are undoubtedly the areas where deep-fakes are doing the most harm.
As for politics more broadly, my instinct is that the concern is a little overblown. Craig spoke to Chipo in this episode about a case in Sri Lanka where an image of a mosque congregation taken before the pandemic was re-circulated in 2020 to claim that Muslims were breaking lockdown rules. If disinformation campaigns are working well enough with poorly doctored images or re-circulated ones, why bother with deepfakes? The man behind @deeptomcruise, spoke to the FT about how complicated it actually is to create such high quality deepfakes: each video is less than a minute long but requires 24 hours of post-production; plus he has worked in visual effects for five years; has access to high-end hardware and the talents of a Tom Cruise impersonator.
I think what we really need to be concerned about is what Aviv Ovadya calls “truth apathy”. Whilst Craig rightfully reminds us that the term “post-truth” is an ‘epistemological cul-de-sac’ (here is an article I wrote about this earlier this year), he makes the important point that the very existence of these technologies in the information eco-system makes people more distrustful of all information. And we can’t forget that sowing confusion and distrust tends to play into the hands of authoritarians. More often than not, accusations of fake news are used to silence dissent - contemporary examples can be found in Malaysia, Hong Kong, India, Mauritius and more.
Eliza is studying the Global Media & Communications MA at SOAS, and holds a BA in English Literature from Cambridge. Eliza worked as a Communications Intern at UNICEF's South Asia office in Kathmandu, and hopes to become a writer on tech and global internet cultures.