Cheap artificial intelligence tools and widespread staffing cuts across major social media networks are making it easier to spread political falsehoods ahead of the US midterm elections. According to tech watchdogs and campaign officials, operatives can now quickly generate and adjust false posts at very low cost until variations slip past automated moderation filters, leaving political campaigns struggling to respond to waves of online deception.
"You can run an information operation now with quality that was impossible in 2016 for super, super cheap," explained Frances Haugen, a former Facebook data engineer and whistleblower. Haugen warned that bad actors can bypass rules simply by tweaking a post slightly each time they generate it. Meanwhile, social media giants have carried out significant layoffs across their trust and safety teams, increasingly replacing human reviewers with automated AI systems and crowdsourced moderation models.
Campaigns across the United States are already dealing with the consequences. In Texas, Democratic Senate candidate James Talarico stepped away from campaign events to recover from the flu, but online provocateurs quickly circulated fabricated claims that he was hospitalized in Ohio. Researchers at the Digital Democracy Institute of the Americas tracked how the lie mutated across conservative Spanish-language media, transforming into a false claim of a suicide attempt. The researchers pointed out that platforms enforce their content rules much less strictly in languages other than English, allowing harmful fabrications to spread with fewer guardrails.
Political candidates also face difficulties finding someone at tech companies to handle these falsehoods when they appear. In addition to fabricated rumors, Talarico was targeted by a campaign advertisement from his Republican opponent, Ken Paxton, featuring AI-generated footage that falsely depicted him rewriting the Bible. Because platforms respond slowly and rely heavily on automated tools, researchers emphasize that standard factual corrections often cannot keep pace with rapidly shifting digital lies.