AI is hurting US politics. Two technologists say it could help instead

AI has mostly made US politics worse so far, Bruce Schneier and Nathan E. Sanders argue in an essay for the Guardian: campaigns use AI deepfakes to spread lies, the White House posts AI-generated "slopaganda," and voters are anxious about what AI means for the country. But the authors say candidates are missing a real opportunity, using AI to listen to voters instead of just broadcasting at them.

AI can help candidates connect one-to-one with as many people as possible. Or it can facilitate many-to-many connections, engaging voters in deliberation about issues at scale.

• Team Mirai (Japan): AI interviewer chatbot for policy input, 300,000+ messages across 16,000 interviews, won 12 seats in the Diet
• CrownShy (Scotland): AI-powered "Comhairle" tool for large-scale constituent consultations, partly funded by the Scottish government
• US tools already in use: Stanford's deliberation.io (piloted by Washington DC), MIT's Cortico (library listening sessions), Talk to the City, and Remesh
• Higher Ground Labs, a Democratic-aligned campaign tech investor, launched a fund this spring targeting "AI-native campaign systems"

Team Mirai calls itself a "utility party," building tools any Japanese political party can use. Its AI interviewer draws out voters' views on specific policy issues through extended conversations, then feeds that input into an AI-powered portal for exploring bills. Members have cited those AI interviews directly in legislative committee hearings, published a synthesis of the input, and amended their policy platform based on it. The party explicitly follows the "gov zero" civic-hacker movement that won political influence in Taiwan.

CrownShy takes a many-to-many approach: its Comhairle tool combines AI interviewing with software for synthesizing viewpoints, running virtual assemblies, and sharing video testimonials, aimed at letting legislatures or campaigns consult constituents at scale. Both Comhairle and Team Mirai's tools are open source. The US has its own version of this civic-tech tradition, dating back to Code for America and the Obama 2008 campaign, which pushed Congress to fund the US Digital Service in the first place. The Trump administration repurposed that agency into the US DOGE Service in 2025.

Government-backed civic AI is having a moment beyond campaigning too. Intokened covered South Korea's own push to give every citizen free access to generative AI through state-picked apps, a different model built for public services rather than elections, but part of the same broader question: who controls the AI that citizens interact with, and on whose terms.

Schneier and Sanders don't treat AI adoption as unconditionally good. They want big tech money out of politics and AI companies held accountable for the harm their models cause. That's exactly the accountability fight already playing out in court. They also want AI company revenues taxed and, if the AI bubble bursts, are open to nationalizing the companies involved.

Their bet is that congressional candidates looking for authentic connection with voters, and a way to stand out from opponents, have reason to try broad listening rather than write AI off entirely.

Nothing here should be taken as financial advice — just information to consider.