Scams using deepfake technology will increase in 2024.

What is Deepfake?

Deepfake is a term coined from the combination of "Deep Learning" and "Fake," referring to a technology that simulates human facial features and creates fake audio, visual, or even video content. Deepfake creators can manipulate media and replace the real person's appearance with that of a real individual.

Essentially, deepfake technology is built on Google's open-source machine learning platform. Deepfakes scan videos and portrait photos of a person, then merge them with separate videos using artificial intelligence (AI) technology, replacing facial details such as eyes, mouth, and nose with realistic facial movements and voice.

The more original images available, the more data the AI ​​has to learn and create more realistic fake images or videos. Deepfakes can superimpose one person's face onto another in a video with astonishing realism.

Deepfakes will increase in the 2024 election year.

The use of deepfakes to impersonate politicians is becoming increasingly common, especially as 2024 is the biggest global election year in history.

According to reports, at least 60 countries and more than 4 billion people worldwide will be electing their leaders and representatives this year, making the issue of deepfakes a serious concern.

According to a report by Sumsub, online media, including social media platforms and digital advertising, recorded the highest increase in identity fraud between 2021 and 2023, at a rate of 274%. Industries such as healthcare, transportation, and video games were also among those affected by identity fraud.

Furthermore, in a 2024 global threat report, cybersecurity technology company CrowdStrike (USA) stated that with a large number of elections approaching this year, misinformation and/or fake news may emerge to sow instability.

Accordingly, prior to the Indonesian elections on February 14th, a video circulated on social media showing the late President Suharto endorsing the political party he once led. However, this video was identified as a deepfake created by AI, which copied his entire face and voice. This deepfake video attracted 4.7 million views on social media platform X (formerly Twitter).

This isn't the first time deepfake videos have surfaced. In Pakistan, a deepfake hoax about former Prime Minister Imran Khan occurred during the election period, claiming his party would boycott the election. Meanwhile, in the US, voters in New Hampshire suddenly read a fake news story about President Joe Biden asking them not to vote in the primary election.

Experts believe that most deepfake products are created by actors within each country. Carol Soon, a researcher and head of the culture and society department at the Singapore Institute of Policy Studies, said that domestic actors could include opposition groups and political rivals, or far-right and left-wing extremists.

Meanwhile, Simon Chesterman, Senior Director of AI Project Governance at Singapore's National AI Programme, said that Asia is not yet ready to address the deepfake issue in elections from three perspectives: legal, technological, and educational.

What are the dangers posed by deepfakes?

Commenting on the dangers posed by deepfakes, Carol Soon stated that, at a minimum, deepfakes pollute the information ecosystem and make it difficult for people to find accurate information and form well-founded opinions about a party or candidate.

In this regard, Simon Chesterman stated that voters may also turn against a particular candidate if they see content about a scandal spreading rapidly on social media before it is exposed as fake news. While some governments have tools to combat online misinformation, the concern is that the truth will be distorted before it can be corrected.

“We’ve seen how quickly X (formerly Twitter) can be taken over by deepfake pornography involving singer Taylor Swift. These things can spread incredibly fast,” Chesterman added, noting that countermeasures are often insufficient and extremely difficult to enforce. “Often it’s too late.”

Adam Meyers, from cybersecurity technology company CrowdStrike, says that deepfakes can also elicit confirmation bias, an effect in human information processing. “Even if they know deep down it’s not true, they’ll still accept it if it’s the message they want and what they want to believe,” he explains.

Adding to these concerns, Chesterman argued that a fake video depicting election misconduct, such as vote-buying, could erode voter confidence in the legitimacy of the election.

According to Ms. Soon, candidates can deny the truth about themselves by claiming that negative information about them is fabricated.

Leading technology companies are committing to efforts to combat deepfakes.

Last February, 20 leading technology companies, including Microsoft, Meta, Google, Amazon, IBM, along with artificial intelligence company OpenAI and social media companies such as Snap, TikTok, and X, agreed on a joint commitment to combat the use of artificial intelligence for fraudulent purposes in this year's election.

The signed commitment is an important first step, but its effectiveness depends on implementation and enforcement. According to Ms. Soon, to achieve this, a multi-faceted approach is needed for technology companies in applying different measures across their platforms. Technology companies also need to ensure transparency regarding the decisions made, such as the processes implemented.

However, according to Simon Chesterman, one shouldn't expect too much from private companies performing basic community functions. Deciding what content is allowed on social media is a difficult decision, and some companies may take months to make a decision.

To achieve this goal, the Content Verification and Validation Alliance (C2PA) has introduced a platform dedicated to verifying digital information for content. This platform allows viewers to receive verified information such as the creator, source, and creation time, as well as the generality of the information or whether it was created by AI.

C2PA member companies include Adobe, Microsoft, Google, and Intel. OpenAI announced earlier this year that it would deploy C2PA authentication technology for images created using the DALL·E 3 image modeling language.

In an interview with Bloomberg House at the World Economic Forum last January, Sam Altman, founder and CEO of OpenAI, said the company is strongly focused on ensuring its technology is not used to manipulate elections.

Adam Meyers, from the cybersecurity technology company CrowdStrike, proposed the creation of a non-profit, non-partisan technical organization tasked with analyzing and identifying deepfakes. “The public could send the organization content they suspect has been tampered with. While this isn’t easy to communicate, at least there would be some mechanism people can rely on,” Meyers said.

Ultimately, while technology is part of the solution, much of it rests with the user. The public needs to be more vigilant; in addition to verifying authenticity when something seems highly suspicious, users also need to verify the authenticity of important information, especially before sharing it with others.