Deepfakers”, the chieftains of the 21st century

Deepfake is the manipulation of digital media (images or videos) to realistically replace a person’s face with another face | Pixabay

Choose your reading level:

STANDARD

Deepfakers”, the chieftains of the 21st century

Written by Laura Casamitjana

Artificial Intelligence (AI) has become a powerful tool for spreading disinformation, particularly in the context of elections. AI algorithms can generate deepfake materials (audio, video, and images) that alter the appearance, voice, or actions of people.

Deepfakes, also known as “synthetic media”, are generated videos, images or audio that mimic the appearance and sound of a person. This technology is so convincing in its simulation of reality that it can fool both humans and the algorithms themselves. It is often used in videos or in elements such as augmented reality filters.

The term deepfake comes from the combination of “deep learning” and “fake”, reflecting the fact that it refers to something fake produced by deep learning technology. There are consumer applications that use this technology for entertainment, such as FaceSwap, its growing availability, popularity and increasing accuracy and resources raise concerns about its potential malicious use.

Manipulating candidates, manipulating the population

United States, 2024. The candidates are taking their positions. Among them, the shadow of Trump looms, backed by his classic conspiracy-minded supporters. This time their target is once again the Democrat, Hillary Clinton: Trumpists are viralising and giving credence to a deepfake showing Clinton endorsing Republican Ron DeSantis. “Donald Trump’s supporters have taken credit for producing a fake video showing Hillary Clinton endorsing Ron DeSantis for president,” said the OECD.AI Policy Observatory, calling the video a hoax.

Deepfaking in the context of political elections has become one of the most important malicious uses of the tool. This is not the first time such a phenomenon has been seen, as disinformation using this method is becoming increasingly common, undermining people’s trust in institutions and providing a wide resource for manipulation. Mario García, General Manager for Spain and Portugal at Check Point Software, cites the example of the Slovakian elections, where deepfake recordings of a party leader posing as a liberal candidate and talking about rigging the elections were circulated on social networks, potentially influencing the results. Using the same technique, an automated phone call was distributed in January 2024 in which the fake voice of Joe Biden, the current US president, and a candidate in the election, was heard urging people in New Hampshire not to vote.

The favourite stage of disinformation

Elections are the peak of deepfake: during elections, there is a notable increase in the use of fake accounts to spread disinformation or fake news, with up to an 80% increase according to Meta (Facebook, Instagram, and WhatsApp). These fraudulent accounts are becoming increasingly difficult to identify for both platforms and users, largely due to the use of artificial intelligence. This is highlighted by Mario García, who explains that the computer security technology company has analysed more than 36 election campaigns worldwide and found that a third of them contain audio and video material generated by AI: “This highlights an evolution towards disinformation and manipulation of democratic processes. The sophistication and accessibility of AI technologies has reached a point where they can easily be used to influence the democratic process”.

Citing a report by Check Point Research, highlights the fact that internet criminal markets are increasingly using sophisticated tools with AI functionality to create deepfakes and manage social media accounts.

A paradigmatic case is a platform offered on a Russian underground forum that uses AI to automate the distribution of content through fake profiles on Instagram and Facebook. This platform can manage hundreds of accounts, post daily, and is particularly adept at large-scale election influence campaigns.

Google, Microsoft, and Meta have recently agreed to limit, but not ban, election deepfakes. In addition to creating fake content, AI is also acting as a coordinator of mass disinformation releases, as Mario Garcia notes: “Social networks play a crucial role in the amplification of false and manipulated content due to their ability to spread information quickly to a large audience.

AI’s yin and yang

With great power comes great responsibility. Artificial intelligence is a powerful resource that has a strong impact and a real ability to influence. Between yin and yang, the misuse of AI for election deepfakes can be countered by the good use of AI for fact-checking. “It is possible to use advanced AI models to build tools to detect false information. Thanks to the characteristics of these models, such as multilingualism or their ability to take into account the context and semantics of language, they can offer solutions where some barriers, such as language, are diluted,” says the CIVIC Project’s report “Fighting disinformation through artificial intelligence“. Commenting on the exponential increase in quality and efficiency that AI brings to fact-checking, the report adds that “these types of tools are and will be a fundamental help to fact-checking organisations, expanding their scope and reach by being able to analyse a greater amount of disinformation circulating on social networks and other media”.

complementary activities

EASY

Deepfakers”, the chieftains of the 21st century

Written by Laura Casamitjana

Artificial Intelligence (AI) has become a powerful tool for spreading disinformation, particularly in the context of elections. AI algorithms can generate deepfake materials (audio, video, and images) that alter the appearance, voice, or actions of people.

Deepfakes, also known as “synthetic media”, are generated videos, images or audio that mimic the appearance and sound of a person. This technology is so convincing in its simulation of reality that it can fool both humans and the algorithms themselves. It is often used in videos or in elements such as augmented reality filters.

The term deepfake comes from the combination of “deep learning” and “fake”, reflecting the fact that it refers to something fake produced by deep learning technology. While there are consumer applications that use this technology for entertainment, such as FaceSwap, its growing availability, popularity and increasing accuracy and resources raise concerns about its potential malicious use.

Manipulating candidates, manipulating the population

In the United States, in 2024, candidates are taking positions for the elections. Among them, the shadow of Trump reappears, supported by his supporters who believe in conspiracy theories. This time, their target is once again Hillary Clinton: Trump’s supporters are spreading and giving credibility to a deepfake showing Clinton endorsing Republican Ron DeSantis. The OECD.AI Policy Observatory has called the video a fake.

The use of deepfakes for political purposes during elections has become an increasingly common problem. Disinformation through this technique undermines trust in institutions and becomes a widespread source of manipulation. Mario Garcia, General Manager for Spain, and Portugal at IT security technology company Check Point Software, mentions that in previous elections in Slovakia, deepfakes circulated on social networks where the leader of one party pretended to be a liberal candidate, potentially influencing the results. In January 2024, a robocall was distributed with the fake voice of Joe Biden, the current US president and candidate for re-election, urging New Hampshire residents not to vote.

The favourite stage of disinformation

The use of deepfakes peaks during elections. There is a significant increase in the use of fake accounts to spread disinformation or fake news, up to 80% more according to Meta (Facebook, Instagram, and WhatsApp). These fraudulent accounts are becoming increasingly difficult for both platforms and users to identify, largely due to the use of artificial intelligence (AI).

Mario Garcia of IT security technology company Check Point Software says they have analysed more than 36 election campaigns around the world and found that a third of them contain AI-generated audio and video material. This highlights a move towards disinformation and manipulation of democratic processes. The sophistication and accessibility of AI technologies has reached a point where they can easily be used to influence the democratic process.

According to a report by Check Point Research, there has been an increase in the use of sophisticated tools incorporating AI in criminal internet markets, particularly for the creation of deepfakes and the management of social media accounts.

One example is a platform offered on a Russian underground forum that uses AI to automate the distribution of content through fake profiles on Instagram and Facebook. This platform can manage hundreds of accounts, post daily, and is particularly adept at large-scale election influence campaigns.

Although Google, Microsoft and Meta recently agreed to limit election deepfakes, they will not ban them. In addition to creating false content, AI acts as a coordinator of mass disinformation campaigns. Social media plays a crucial role in amplifying false and manipulated content because of its ability to spread information quickly to a large audience.

AI’s yin and yang

With great power comes great responsibility. Artificial intelligence is a powerful resource that has a strong impact and a real ability to influence. Between yin and yang, the misuse of AI for election deepfakes can be countered by the good use of AI for fact-checking. “It is possible to use advanced AI models to build tools to detect false information. Thanks to the characteristics of these models, such as multilingualism or their ability to take into account the context and semantics of language, they can offer solutions where some barriers, such as language, are diluted,” says the CIVIC Project’s report “Fighting disinformation through artificial intelligence“. Commenting on the exponential increase in quality and efficiency that AI brings to fact-checking, the report adds that “these types of tools are and will be a fundamental help to fact-checking organisations, expanding their scope and reach by being able to analyse a greater amount of disinformation circulating on social networks and other media”.

☑️ Test your knowledge

Reading Comprehension questions. Deepfakers”, the chieftains of the 21st century

Step 1 of 3

According to the text, what are the risks of using deepfakes in electoral processes?
Skip to content