Deepfakes: Fun, art, or manipulation?

AI technology has become increasingly apparent in our society, taking huge strides in record time. Photos, videos, and audio clips manipulated or created by AI fall under the title of ‘deepfakes,’ and can be deeply misleading to their audience. But is there any cause for real concern?

Deepfakes are the product of AI manipulating content to look real, either transforming someone into someone else or even generating a fake version of them. This includes photos and videos, some even with mimicked voices.

Greater use and sharing of the products of deepfake AI across social media mean that the technology behind it works rapidly to improve itself

According to a study conducted in 2017 by V Schetinger et al, 53.5% of humans could be tricked into believing that digitally altered media is real. Years later, despite the impressive technological development and increase in our understanding of AI, the assumption that this figure will have decreased is not a clear one. Greater use and sharing of the products of deepfake AI across social media mean that the technology behind it works rapidly to improve itself and the quality of its products, creating increasingly realistic work.

Between 2019 and 2020, the amount of deepfake content online increased by an astounding 900%, according to the World Economic Forum. This sets the tone for its trajectory in the years to come. With such a large online presence, it leaves us wondering how it truly impacts our lives.

Many famous examples appear to show deepfake material as merely harmless fun. The AI-generated image of Pope Francis embracing Madonna, which she posted on Instagram in December 2024, did cause some controversy. The Pope is a popular choice for deepfakes, due to the plethora of online videos and photos. The greater the online presence, the greater the target they become for the creation of AI deepfakes. The Madonna image was created using a paid AI platform, Krea.ai, but you can find realistic deepfake images of many celebrities on free-to-use platforms with ease.

We must understand how these new possibilities can lead to greater mistrust in the media

Being such a versatile tool, perhaps the danger and the weight of its influence have been overlooked, especially when it comes to politics and scams. Yet it has also become a means of creating art for some users. Could it be a process of advancing the art scene in the digital age? In one sense, many are using these new platforms to develop their imagination and creativity, portraying humorous or satirical ideas with the aid of new technology. On the other hand, we must be cautious of harmful uses of these images, as they can be easily manipulated and transformed into photos of celebrities or even politicians in positions of immense power, and could be used to ‘cancel’ such individuals, despite a lack of awareness of their true authenticity. Even if this content was initially created with comedic intent behind it, depending on who is chosen as a target, deepfakes could have severe consequences.

There is such a high volume of AI-generated content on social media, and eventually (if it is not already happening) people will struggle to tell the real from the fabricated. We must understand how these new possibilities can lead to greater mistrust in the media, the spreading of misinformation, and more scams taking advantage of those vulnerable to this type of content.

A recent example of this took place in 2024, involving a 77-year-old retired lecturer from Edinburgh. The BBC explains how she had been convinced by scammers using deepfake AI-manipulated videos that she was in a relationship with a woman working on an oil rig. Using videos, the scammers convinced her that their AI-generated woman was a real person, giving the victim the confidence to send them money so that they could continue talking, reaching a total of roughly £17,000. Despite her age arguably pinning her as more susceptible to this type of scam, it is vital to see the possibility of this happening to those who do not consider themselves to be as vulnerable – the rapid development of AI may well be beneficial, but in the long run, it could serve to work negatively against us in scam scenarios.

Lower quality deepfakes are particularly obvious due to unnatural movement or robotic voices, however, the BBC has shared other tips on how to identify AI

So far, there are no easy ways to automatically and accurately identify deepfake content from the real. However, there is a general understanding of the techniques used. ‘Face swapping’ and the mapping of voices onto generated videos are only two examples of many. Yet this can help us determine for ourselves the fake from the real if there comes a time when it is unclear. Lower quality deepfakes are particularly obvious due to unnatural movement or robotic voices, however, the BBC has shared other tips on how to identify AI. These include checking the number of fingers on the person depicted, as AI can give itself away when they have more or fewer than five. Also included is an encouragement to look at the eyes of the person depicted – both placement and movement is very telling – and body language too can appear jerky or unnatural. Finally, check the background to see if it matches the supposed time and place when the video or photo was taken.

In general, it is a good idea to be wary of online material that we come across, since fake news and deepfakes are branches from the same tree with roots in misinformation and fearmongering. Whilst also being harmless in some scenarios, it is important to be aware of the potential threat this type of content poses to avoid becoming a victim. However, that is not to say that people should not enjoy the creative aspects of this AI deepfake technology, especially as a dive into satirical art or social commentary, as it can be an incredibly effective tool in getting messages across to a wide audience in an easily digestible format.