AI Disinformation Targets US Midterm Candidates on Social Media

AI-driven disinformation is proliferating on social media platforms ahead of the US midterm elections, enabled by high-quality synthetic tools and reduced content moderation. These tools allow for the cheap production of fabrications that can be slightly modified to bypass automated filters.

In Texas, Democratic candidate James Talarico has been targeted by false claims that he was hospitalized in a mental health facility and AI-generated ads depicting him rewriting the Bible. In Pennsylvania, the U.S. Department of Homeland Security amplified a false claim that an undocumented immigrant served as a campaign worker for Democrat Janelle Stelson. Tim Wallach, Deputy Director of Immigration and Customs Enforcement, stated it was unbelievable that Stelson would employ an illegal alien.

Major platforms including Meta, Google, and X maintain policies against misleading voters about voting logistics but have narrowed restrictions on political commentary. These companies have shifted toward AI-led or community-based moderation, such as X's crowdsourced community notes program. Critics argue these shifts allow disinformation to slip through the cracks despite requirements for disclosing synthetic images in election ads.