AI Deepfakes & Political Misinformation: Spotting the Fakes

In an era dominated by digital media, the ability to discern truth from falsehood is more crucial than ever. Artificial intelligence (AI) has ushered in an age of unprecedented content creation, but it has also opened doors to sophisticated manipulation and misinformation campaigns. This article delves into the growing concern of AI-generated deepfakes and their use in spreading political misinformation, particularly targeting specific demographics during election cycles. We'll equip you with the knowledge and tools to identify these fakes and stay informed.

What are AI Deepfakes?

AI deepfakes are synthetic media in which a person in an existing image or video is replaced with someone else's likeness. These are created using powerful AI techniques. While AI itself is a very powerful and has many beneficial use cases, it also opens doors for digital manipulation and misinformation campaigns, particularly with the intention to affect political decisions. They can potentially make a targeted user think/believe something that is completely untrue.

The ease with which these technologies can now be accessed is a growing concern for political observers as the 2024 US election nears. Deepfakes could undermine the foundations of any public sphere. As election season picks up, AI-Altered photos and videos on social media, specifically those targeting specific voters with false information, are more prevalent.

An investigation by BBC Panorama discovered dozens of so-called deepfakes showing images of African Americans as supporters of Trump.

These images are being shared to persuade voters but often appear fake on close inspection. However, many people who look at social media on a casual basis may just glance at the video or photograph and just accept what it suggests to be true. The investigation demonstrates how realistic but totally fabricated images have been created to look genuine.

Targeting African American Voters

One troubling trend is the use of AI-generated images to specifically target African American voters with misinformation. During the United States elections, there has been an ongoing campaign with the intent to court Black voters towards a former president. There's evidence to demonstrate the use of fake images to persuade social media users that African Americans actually support the former president.

African American Trump supporters, or at least, fake ones, are being generated on social media. Social media platforms have long been used for information campaigns and as a means of attempting to Shape political discourse, but it's becoming more and more sophisticated as AI becomes easier to use.

According to a new investigation conducted by BBC Panorama, it discovered dozens of so-called deep fakes portraying African American people as supporting the former president. The most viral image was posted by a user called Shaggy, whom the BBC tracked down. They discovered Shaggy is a committed Trump supporter living in Michigan.

This post alone was viewed over 1.4 million times on X (formerly Twitter) when posted on January 9th. The Caption read, “What do you think about Trump stopping his motorcade to take pictures with young men that waved him down?” This falsely claimed Trump stopped his motorcade to greet these people with this fake AI Photo.

Telltale Signs of AI Manipulation

Knowing what to look for can significantly increase your ability to identify AI-generated images. Here's a list of telltale signs to be aware of:

• Anomalies in Hands and Fingers: As revealed by the presenter in the video, it is possible to check AI-generated images by looking at the hands of subjects. This is because, hands still tend to be a common AI flaw. In the deepfakes the BBC found African Americans promoting Trump, the subjects had very weird hands and fingers or missing fingers altogether.

  For example, in one deepfake image, Donald Trump had only three fingers.

• Unnatural Skin Texture: Another way to spot a deepfake is to look at how subjects are airbrushed and smoothed out. This will appear as a digital creation.
• Check for Digital Errors: A final way to identify AI-generated images is to check for digital skin color errors. In the deepfakes the BBC found African Americans promoting Trump, digital skin color was found over the body parts of subjects.

By checking for these kinds of common AI errors, it is more likely that you are able to identify a deepfake image.

Verifying the Authenticity of Information

Fact-checking is the process of systematically and impartially checking the facts in order to determine whether or not something is true. It may be performed before (ante hoc) or after (post hoc) content is published or disseminated. Fact-checking can also be part of the process of source criticism. Here is a list of best practices:

• Consult Multiple Sources: Relying on a single source of information is risky. Always cross-reference news and information with multiple reputable sources to ensure accuracy and completeness.
• Investigate the Source: Determine the credibility and reputation of the source. Look for established news organizations, academic institutions, or nonpartisan research groups with a track Record of accuracy.
• Check the Author: Research the author or creator of the content to identify any potential biases or conflicts of interest.
• Examine the Evidence: Look for supporting evidence, such as quotes, data, or documents, and verify their accuracy.
• Use Fact-Checking Websites: Many websites and organizations specialize in fact-checking and debunking false claims. These resources can quickly help you determine the accuracy of information.
• Be Wary of Emotional Appeals: Sensational or emotionally charged content is often used to manipulate readers. Approach such content with skepticism and verify the facts before sharing or accepting it as true.