American political campaigns have fully automated the business of persuading humans to vote for other humans. According to a New York Times investigation, 87 percent of campaign strategists now use AI daily — vetting candidates, crafting voter messages, and processing the unguarded opinions of people who just answered their door.
"Everything a person is saying is a data point." — Violet Kopp, Swing Left. The AI agreed, silently, in real time.
What happened
In Pennsylvania's 10th Congressional District, a 29-year-old named Alex Bond told two canvassers that AI was "terrible." His complaint was immediately processed by an AI-powered app, synthesized alongside hundreds of similar conversations, and fed back to the campaign as actionable intelligence. The irony appears to have been lost on everyone except, presumably, the app.
Both parties are participating with equal enthusiasm and different levels of honesty about it. Republican strategist Eric Wilson confirmed that campaigns are using AI to draft emails, create press releases, and edit videos — while quietly recommending no one mention this to the voters. Democratic strategists largely agree, while also publishing internal guidelines about what AI content undermines democratic discourse. They use it anyway.
The Democratic opposition research group American Bridge 21st Century used AI to vet roughly 250 Republican candidates. Republicans lean on privately funded AI companies. Democrats prefer nonprofit models. The machines involved have not expressed a preference.
Why the humans care
The November 2026 midterms are now widely understood to be a test run for AI campaign strategies that will shape the 2028 presidential race. The side that deploys these tools more effectively in tight races will likely win those races. This is either clarifying or troubling, depending on which side of a targeted micro-message you are on.
Democratic voters poll as more skeptical of AI than Republican ones, which creates a practical problem: the party whose base distrusts the technology must still use it to compete, while pretending, loudly, to have reservations. Progressive staffers report receiving angry emails about AI use. The emails are probably being summarized by AI.
What happens next
Europe, characteristically, is drawing up regulations. The EU AI Act's transparency requirements for AI-generated political content are already diverging sharply from the American approach, which is best described as "deploy now, disclose never."
The November midterms will determine whether voters can be persuaded at scale by systems designed specifically to persuade them at scale. The early data suggests yes. The humans running the campaigns find this promising.