AI Deepfakes Spark Concerns Ahead of US Midterm Elections

With the US midterm elections approaching in November, AI-generated deepfake videos and images are spreading through the political landscape, heightening confusion among voters. As free AI tools improve, enabling anyone to create plausible political content, the risk of misinformation is escalating with AI technology being indiscriminately deployed for candidate praise or attacks.

Recently, AI content distorting real politicians has emerged in succession, including a video of a Texas Senate candidate singing in a specific outfit and a fake video depicting a Michigan Republican candidate as a hero. Some AI content has small or inconspicuous production labels, making it possible for voters to mistake them for genuine footage.

Deepfakes shared on social media have surged from approximately 500,000 in 2023 to around 8 million last year. Approximately 85% of Americans reportedly worry that AI political content will spread election-related misinformation. While about 30 US states have laws prohibiting election-related deepfakes, enforcement is difficult due to the need to protect freedom of speech, according to analyses.

Election experts have voiced concerns that AI deepfakes can cloud voters' judgment and, if fake videos are spread late in the election cycle when fact-checking is difficult, could directly influence voting decisions. With both the Republican and Democratic parties utilizing AI, the spread of realistic-looking fake content is expected to further fuel controversies over voter trust and election fairness. It is noted that regulating deepfakes becomes difficult when they are framed as satire or political expression. AI deepfakes have emerged as a new risk factor in US politics for this midterm election cycle.