The election campaign also strives for "AI recommendations!" US candidates adjust their propaganda content to make AI answers more relevant to their political opinions

The U.S. midterm elections are approaching, and the election battleground extends from social media and search engines to artificial intelligence (AI) chatbots. As more and more voters use AI chatbots to ask candidates about their policy positions, some candidates are also adjusting their campaign propaganda content based on AI large-scale language models, hoping that AI's answers will show them as a better choice.

Reuters exclusively reported on the 26th that a campaign strategy brief prepared for Mike Minogue, the Republican candidate running for governor of Massachusetts, aims to let ChatGPT, Claude, Gemini and other AI chatbots present Minogue as a better choice on issues such as taxes, housing, and public safety. The brief recommends that the campaign team build hundreds of policy pages that are easy for AI to read, and create content on AI frequently cited websites for AI to quote. At the same time, based on how the main AI chatbot responds to political questions, the content is systematically modified until the AI’s answers better reflect the candidates’ political views.

This approach has evolved from the "search engine optimization" (SEO) commonly used in traditional campaigns to the so-called "generative engine optimization" (GEO). As AI becomes more and more the channel through which people obtain news and information, political campaign teams must become more proactive in influencing the answers generated by AI so that AI can recommend their candidates when voters ask them about their performance on some election issues. The core concept is that campaigns not only directly persuade voters, but also gradually shape the information environment by providing data on the Internet that can be referenced by AI, affecting how AI describes, compares and recommends candidates.

The Texas company Visigent, which developed the aforementioned strategy for Minogue, counts six political campaigns among its clients. CEO Thomas Howell said that if you don't put "correct information" on the Internet, "who knows what content the AI ​​will capture." The company's briefing provided to Minogue's campaign said that campaigns now have to compete to be "recommended objects in AI answers."

Joe Bonilla, a partner at political strategy firm Relentless Awareness, predicts that “campaign campaigns utilizing AI influence are now a reality and will become ubiquitous by the 2028 presidential election cycle.”

However, experts are concerned about fact-checking and misinformation in this type of AI campaign. Meg Schwenzfeier, an analyst at Caucus AI who tracks voters' use of AI, said chatbots tend to have a very positive tone but may absorb a large amount of unverified information, so they must beware of false information. The Minogue campaign did not respond to inquiries from Reuters.