The spread of synthetic media on X

Abstract

Generative artificial intelligence (AI) models have introduced new complexities and risks to information environments, as synthetic media may facilitate the spread of misinformation and erode public trust. This study examines the prevalence and characteristics of synthetic media on social media platform X from December 2022 to September 2023. Leveraging crowdsourced annotations identifying synthetic content, our analysis reveals an increase in AI-generated media over time, with an initial spike in March 2023, following the release of Midjourney V5. While most synthetic media identified is non-political and non-malicious, concerning deepfakes targeting political figures persist, raising questions on the potential for misuse of AI technologies.