The survey, performed by U.K. research firm Savanta, asked respondents to assess how confident they were in identifying deepfakes, which are videos or images that have been altered with artificial intelligence or editing software to create a false impression, according to the Guardian.

The survey reported that 45% of Republican voters were either only slightly confident or not confident at all in their ability to recognize fake images or video, as opposed to Democrats’ 35%. Republicans were also more likely than Democrats to say public figures should clearly label when they use AI or manipulate images or videos, with Republicans at 72% and Democrats at 66%.

A majority of people from both parties, 76%, said more should be done to protect the public from false images.

“This is a very real problem and one that voters want social media companies to grapple with,” research analyst Ethan Granholm said, according to the report.

The survey was created in response to the controversy surrounding fake images of Taylor Swift on the social media platform X, which included pornographic material, videos, and images depicting Swift supporting former President Donald Trump’s reelection campaign.

The survey was conducted before the Trump campaign was in the news for using AI images on social media, including depicting Vice President Kamala Harris as a communist dictator. Trump has also claimed the Harris campaign “AI’d” her crowd sizes and has been joined by X owner Elon Musk in spreading false images.

One in four respondents to the survey said they were “disappointed” in Trump sharing fake images of Swift, including 23% of Republicans.

The survey polled 2,004 U.S. adults and was conducted between Aug. 22 and 24.