Politicians Turn to AI Tuning to Shape Voter Perceptions

Candidates are now editing their digital footprints to steer the answers generated by large-language-model chatbots such as OpenAI’s ChatGPT and Google’s Gemini. Democratic primary hopeful Dustin Lloyd of Missouri added a detailed Q&A to his campaign site after a CampSight report flagged his Wikipedia, Ballotpedia and website as “lackluster.” Once the changes were live, the chatbot shifted from recommending his opponent to highlighting Lloyd’s “strongest and most explicit small-business platform.” Lloyd says he is “constantly working on it every day.”

The Straits Times

The practice, dubbed answer engine optimisation, emerged as chatbots began pulling real-time data from the web. CampSight, launched in June by the progressive Run for Something Action Fund, advises candidates to refresh their own sites, update encyclopedic entries, and even post on Reddit to seed chatbot responses. Caucus AI estimates that at least 16 million voters already receive election information from AI, and its own test showed newly edited Wikipedia content can appear in chatbot answers within about 12 minutes. An experiment before Scotland’s parliamentary election in May found more than a third of AI-generated answers were inaccurate, illustrating the speed and volatility of the medium.

The Straits Times

Run for Something’s CTO Jordan Haines notes that “everyone’s just grasping for some sort of tool to at least inspect these results and ideally influence them.” Meg Schwenzfeier, founder of Caucus AI and former chief analytics officer for Kamala Harris’ campaign, warns that as chatbot reliability improves, “there will be a significant uptick in 2028 and beyond.” Tim Chambers, head of digital strategy at Dewey Square Group, cautions that “disinformation actors are, by definition, early adopters – they will use every technique and every technology they can.”

The Straits Times

Experts highlight the danger of AI-generated misinformation. The Scotland study revealed fabricated candidates, false nepotism claims and invented financial scandals. Researchers stress that the underlying models are “black boxes,” making it hard to control or audit the information they surface. The rapid, unvetted insertion of content raises concerns that foreign influence operatives could manipulate voter perceptions as easily as domestic campaigns.

The Straits Times

• “I’m constantly working on it every day.” — Dustin Lloyd, Democratic primary candidate

The Straits Times

• “There is a certain degree of panic around it, because we understand it so little and it’s changing so fast,” — Beth Simone Noveck, professor, Northeastern University

The Straits Times

• “Everyone’s just grasping for some sort of tool to at least inspect these results and ideally influence them,” — Jordan Haines, chief technology officer, Run for Something

The Straits Times

• “As people get more used to this, and as the reliability of the information coming from these chatbots increases, we do think that there will be a significant uptick in 2028 and beyond,” — Meg Schwenzfeier, founder, Caucus AI

The Straits Times

• “Disinformation actors are, by definition, early adopters – they will use every technique and every technology they can,” — Tim Chambers, digital strategist, Dewey Square Group

The Straits Times