How AP reports on elections in the age of AI

Though the impact of AI deepfakes tied to elections hasn’t been as severe as some feared, a wave of misleading AI-generated content did hit elections early this year in Europe and Asia. It served as a warning for countries with upcoming elections, including the United States.

“You don’t need to look far to see some people … being clearly confused as to whether something is real or not,” said Henry Ajder, a leading expert in generative AI based in Cambridge, England.

The question is no longer whether AI deepfakes could affect elections, but how influential they will be, said Ajder, who runs a consulting firm called Latent Space Advisory.

As the U.S. presidential race heats up, FBI Director Christopher Wray has warned about the growing threat, saying generative AI makes it easy for “foreign adversaries to engage in malign influence.”

TIPS FOR IDENTIFYING AI-GENERATED CONTENT

• Try to find the original content creator for confirmation.
• Run reverse image searches on visual content to try to identify the original version.
• If the visual content claims to show a specific location, cross-reference it with authentic photos and videos of the location.
• Look for visual abnormalities, but don’t assume any image that looks weird is AI generated.
• Generative AI tools are improving, so don’t rely on obvious visual abnormalities.
• Contact digital forensics experts who can assess content for indicators that it is fake.
• Use your journalistic common sense: Look for corroborating evidence of the video, image or audio in question. Does the context make sense?

KEY TERMS TO KNOW

The AP Stylebook offers guidance for covering AI (subscription required). Here are some other terms to define for your audience.

• AI: A branch of computer technologies that process data and complete tasks, traditionally using predetermined rules.
• Generative AI: A type of AI model that can create something new. These systems can make realistic text, images, videos or audio recordings based on an inputted prompt.
• Deepfake: A convincing image, video or audio hoax created using generative AI technology, often to impersonate a person or portray an event that never happened.
• Chatbot: A computerized system that simulates a conversation with a human user. Many newly popular chatbots use generative AI to generate readable text on demand and even produce novel images and video. They are based on large language models, a type of deep learning algorithm that has been trained on a large database of digital data.
• Hallucination: A problem with still-evolving generative AI technology in which large language models tend to make statements that sound convincing but are false or made up.