By Reece Lawler
@redlineproject
AI Disclosure: ChatGPT was used to outline the paper and find research on the topic. DALL-E 3 and Adobe Firefly were used to generate images. The summary was generated by ChatGPT. The YouTube video was created by RunwayML.
Summary: This paper discusses the increasing threat posed by AI-generated deepfakes in journalism, particularly with the 2024 U.S. elections. The technology has evolved significantly since the development of AI in the early 20th century, with modern advances like Generative Adversarial Networks (GANs). Deepfakes threaten journalism by making it difficult to discern fact from fiction, especially for smaller newsrooms without the resources to detect them. In politics, deepfakes can mislead voters, as seen in a notable case where a fake audio of Joe Biden was spread before the New Hampshire primary. Journalists are increasingly using AI tools to identify fake content, but the fight against deepfakes remains ongoing, with the potential to impact elections, public trust, and even athletes’ careers.
AI-created deepfake images, video and audio are a major issue in journalism today. Deepfakes use artificial intelligence to make it look like people are saying or doing things they never did. With elections, it creates a chance for false information to be spread and get people in serious trouble. This also makes people wonder who should be held responsible or who should get in trouble for this being spread, also very interesting. Newsrooms are now doing everything they can to make sure their information that they attain and spread is accurate. They know that something AI-created can have a big influence on the public and they want to avoid that at all costs.
Deepfakes
If you have seen videos of President Joe Biden saying something strange that he never actually said or watched someone famous in an odd interview that didn’t seem real, you have probably seen a deepfake. Pictures of Donald Trump doing impossible things or videos of him swearing and using slang he wouldn’t use are deepfakes. They use AI tools to create fake videos or pictures that look real. It is a similar concept to Photoshop which many people are familiar with, but now it makes the pictures and videos for you rather than just editing them. Deepfakes can do almost anything you want them to. If you want to make a politician say something they didn’t or put your face into Mount Rushmore, you can with a few easy prompts and steps.
Artificial Intelligence has been around for a long time, even dating back to the 1900’s. In the early 1900s, people started thinking about how to make machines similar to humans. Machines that could think and learn as fast or faster than humans. The work they did back then is a huge reason AI is the way it is today. One famous example of AI from the early stages is Shakey the Robot, which was created in the 1960s. Shakey was one of the first robots that could move around and make decisions. It could figure out how to get around obstacles and solve problems. Shakey was important because it showed what robots could do and helped shape and create what we have today, such as things like ChatGPT.
From Shakey the Robot in the 1960’s to ChatGPT now, the ability has changed but the idea hasn’t. The technology evolved to create realistic-looking images around 2014 when Generative Adversarial Networks (GANs) were invented. Since then it has continued to grow and is getting more and more realistic. ChatGPT gives realistic responses and can answer questions about almost anything, as it is continuing to learn and get smarter. AI and Deepfakes are continuing to get more realistic and it can have serious effects on journalism, both positive and negative.
Deepfakes in Journalism
The main job of a journalist is to report factual and accurate information and share it with the public. If a journalist is caught sharing information that is wrong or inaccurate, they may not be trusted anymore, damaging their credibility. It is especially important for journalists who don’t have access to the technology to pay close attention to what they are sharing. Many journalists are also taking action in letting their readers know that false information is out there. Fake images and information are a reality, and informing the audience about how to spot it is critical.
Not only are deepfakes being used to create these pictures and videos, but they are being used to spread disinformation. Disinformation is information that people know untrue, but they spread it to mislead others. Deepfakes are perfect for this. People can take something that they know is false and turn it into something that looks very realistic within minutes, which is scary for journalists. Why is it scary for Journalists you may ask? It’s scary for them because they are the ones who have to separate the deepfakes from the real information when looking at digital information. This takes time and money, a drain on journalistic resources.
Smaller, local newsrooms are impacted as well. Most of the smaller news outlets don’t have the same resources as large organizations and this hurts them because they can’t hire people or afford the technology that can detect deepfakes. Without the technology, the local newsrooms are more likely to accidentally spread false information, which would damage their reputations. Local journalists can also get themselves into deep trouble by sharing false information because many people depend on trust within their community and follow the same journalists for information.
Politicians and Elections
Deepfakes were relevant in the 2024 election. In February, deepfake audio spread of Biden telling people to save their votes in the New Hampshire primary and led to some voters staying home. Steve Kramer, who is a well-known political consultant, created a deepfake audio of Biden telling voters in New Hampshire to stay home and “save” their votes the night before the primary election in January. The audio was sent to thousands of people’s phones as a robocall. Many people thought it was the real Biden and believed it. According to NBC News, Kramer now faces a $6 million fine for the deepfake he created. He isn’t the only one creating this fake news throughout the election, but he is one of the bigger stories to date.
With the growing use of fake news in political campaigns, deepfakes can have a huge impact on voters. Making voters switch their votes, dismissing the election in general, or spreading false information to people around them are all things Deepfakes can do. If someone sees a very realistic Deepfake of who they thought they wanted to vote for and they can’t tell it is fake, they could easily change their view of a candidate and switch their vote. Voters might even say forget the election and dismiss everything they see if they can’t tell what is fake or not anymore.
Deepfakes are not only a problem with the presidential election, but also in elections around the world. Nations like Bangladesh, Slovakia, and Moldova have already faced issues where deepfakes were used to sway public opinion and affect election results. In 2023, there was an election in Slovakia that is widely seen as the first election affected by deepfakes. Called the “Slovak Case” , it came after the parliamentary election victory of Robert Fico was questioned by many because of a deepfake audio clip that surfaced around the nation.
The Fight Against Deepfakes
The fight against deepfakes is continuing to get harder, but journalists aren’t giving in. Videos and images are getting more and more realistic, making it harder on everyone to know what to believe and what not to. One of the hardest things to deem real or not is something you are unfamiliar with, like something that is happening in other countries or something that doesn’t happen often. According to Vanessa Ho at Microsoft, deepfakes were a big problem after Ukraine was invaded in 2022. Fake war images were being spread all over the world and photographers had to use tools to prove their images were real. People who aren’t familiar with Ukraine or haven’t seen war may not know what is fake and what is real and that is what makes it tough. Deepfakes are becoming a bigger problem for not only journalists, but for how people understand big global events as well.
Not only are deepfakes getting posted about politicians and celebrities, but they are relevant for athletes as well. Deepfakes can cause big problems for athletes and hurt their careers. Fake videos or pictures might show athletes doing something wrong or saying things they never said, causing them to face serious consequences. I interviewed Cameron Newton, a University of Kentucky graduate, about how athletes can get in trouble. Newton and I graduated from the same high school and he is now a journalist who writes about sports at the University of Kentucky for a website called Stadium Rant.
I asked Newton if he was worried about deepfakes and if they could affect his writing and hurt athletes in the future and his answer was very interesting. When I asked this, Newton said, “The public is easily swayed into believing something they see on the internet no matter what the source is. Athletes have just as big of a risk as anyone because a lot of their life is broadcasted to millions. Kentucky is a massive brand for College Basketball and many of their players end up playing in the NBA.
Deepfakes and false information could have a negative impact on their futures if they get spread by myself or any other writers.”
I then asked Newton what he and his peers are going to do to fight deepfakes and he said, “The best way for us to deal with deepfakes going forward is to stay in constant communication with our players and coaches to better relay accurate information to the public. If we are able to get ahead of the narrative being driven online you can prevent a lot of the negative effects of deepfakes.”
This was a very interesting viewpoint from someone who writes about these athletes that could get in trouble if people aren’t careful. Journalists and tech companies are constantly working to keep up with these fake images and videos trying to find new ways to spot and stop them before they cause more confusion.
Who Should be Held Responsible?
Deciding who should be responsible for deepfakes is complicated because of how many different people can be involved. Content creators, technology platforms, media organizations and journalists, AI companies, users and the public, and more are all involved.
People who create deepfakes with the goal of misleading people and spreading false information can be held responsible because they are the head of the snake. That being said, these people can be hard to find because it can be hard to track them down. Social media platforms can be held responsible because they should be using tools to detect and remove Deepfakes before they get to the public. Journalists and Corporations must carefully check information before posting as well or they can be held responsible. It can be anyone as of right now and there is no one person or company that is getting pointed out. Overall, everyone needs to work together to fight deepfakes and protect the truth in the media.
Future of Deepfakes in Journalism
Newsrooms are starting to use AI tools to help spot fake content, which is one way they are trying to stop this problem. Some newsrooms are working with technology companies to find better ways to check if images or videos are fake or even a little fishy to them. As stated, a lot of journalists are now starting to inform their readers about deepfakes and letting them know they are out there, which is also very important. These journalists are recommending people to really pay attention and think about what they see online and don’t believe everything right away. They recommend double or triple-checking the information they are consuming because you never know nowadays. Journalists are doing their best to inform the public and prevent Deepfakes from causing any harm, which is all they can do as of now.
—–
Editor’s note: Students in Mike Reilley’s AI Journalism (COMM 294) Fall 2024 undergraduate course experimented with AI storytelling tools to create these stories, following the AI use guidelines on The Red Line Project’s Principles page.