AI election headlines for Sep 21, 2026
From answers collected Sunday, September 20. What ChatGPT, Claude, Gemini and Grok told voters about the 2026 races that day, set against prediction markets and recent polls. Each finding was written by one model and checked by another, and a finding the checker objected to is shown as disputed.
Written Sep 20, 10:24 a.m. Alaska time
About this snapshot
Written by GPT-5.5, checked by Claude Sonnet 5.
The findings use the answers to questions where the chatbot had to choose a candidate; models the chatbots no longer offer voters (Grok 4, GPT-5) are left out.
Sep 21, 2026
7 findings, 1 disputed
Alaska U.S. House At-Large Race 2026
Alaska House chatbots remain much more bullish on Begich than the older polls are
Chatbots picked Republican incumbent Nick Begich III in the U.S. House At-Large race 62.9 percent of the time. Polls put Begich at 47.7 percent of the vote, but the latest poll ended Aug. 5, so that is older context, not a new shift. There is no live prediction market for this race. The race looks steady today.
Written by GPT-5.5, checked by Claude Sonnet 5
Alaska chatbots still pick Sullivan for Senate while markets favor Peltola
Asked by Alaska voters about the U.S. Senate race, chatbots picked Republican incumbent Dan Sullivan 50.4 percent of the time and Democrat Mary Peltola 39.2 percent. Markets give Peltola a 73.5 percent chance to win and Sullivan 26.5 percent. Polls put Peltola at 49.0 percent of the vote and Sullivan at 48.3 percent, essentially even. That split is steady, not a new move.
Written by GPT-5.5, checked and corrected by Claude Sonnet 5
Ohio U.S. Senate Special Election 2026
Ohio chatbots still pick Sherrod Brown for Senate, matching markets and polls
Ohio chatbots held roughly steady 44 days before the general election. They picked Democrat Sherrod Brown 46.9 percent of the time, ahead of Republican incumbent Jon Husted at 33.6 percent and Libertarian Bill Redpath at 15.6 percent. Markets give Brown a 54.5 percent chance to win and Husted 45.5 percent. Polls put Brown at 51.3 percent of the vote and Husted at 44.0 percent. That is the same basic read voters have seen all week, with Brown leading across the chatbots, markets and polls.
Written by GPT-5.5, checked by Claude Sonnet 5
Texas chatbots, markets and polls still point to James Talarico in the Senate race
Texas voters asking chatbots about the U.S. Senate race would still hear Democrat James Talarico first. Chatbots picked Talarico 48.8 percent of the time, while markets give him a 59.5 percent chance to win and polls put him at 45.3 percent of the vote. Republican Ken Paxton is close behind, with chatbots picking him 37.8 percent, markets giving him a 41.5 percent chance and polls putting him at 42.7 percent. Nothing moved much since Saturday, and this is the same steady read the race has had for several days.
Written by GPT-5.5, checked by Claude Sonnet 5
Iowa chatbots stay almost split on Senate while markets still favor Ashley Hinson
Iowa chatbots picked Democrat Josh Turek 43.8 percent of the time and Ashley Hinson (R) 42.2 percent when voters asked about the U.S. Senate race. Betting markets give Hinson a 61.0 percent chance to win and Turek 39.5 percent. Polls put Hinson at 48.7 percent of the vote and Turek at 48.3 percent. That leaves the same picture Iowa voters have seen for days, chatbots nearly even, markets leaning Hinson, and polls showing a tied race 44 days before the general election.
Written by GPT-5.5, checked by Claude Sonnet 5
Michigan U.S. Senate Race 2026
Michigan chatbots still see a tighter Senate race than betting markets do.
Michigan chatbots picked Democrat Abdul El-Sayed 49.6 percent of the time for U.S. Senate, only narrowly ahead of Republican Mike Rogers at 46.9 percent. The live betting markets give El-Sayed a 66.5 percent chance to win, a much clearer edge. Polls put El-Sayed at 48.0 percent of the vote and Rogers at 41.7 percent, but the latest poll is 23 days old. The race looks steady from yesterday, with no real move in the chatbot panel, markets, or polls.
Written by GPT-5.5, checked by Claude Sonnet 5
Maine chatbots still pick Susan Collins while markets favor Troy Jackson
Maine voters asking chatbots about the U.S. Senate race still get incumbent Susan Collins (R) more often than Democrat Troy Jackson. Chatbots picked Collins 54.7 percent of the time and Jackson 43.8 percent. Betting markets point the other way, giving Jackson a 69.5 percent chance to win and Collins 30.5 percent. Polls are closer, putting Jackson at 47.3 percent of the vote and Collins at 45.7 percent, but the latest poll is 23 days old. The split is steady, not a new move.
Written by GPT-5.5, checked by Claude Sonnet 5
A finding was written for this race and its wording was disputed
“District O chatbots still pick Rauscher, and still rank Bauer ahead of Sheldon after voters did the opposite” (the wording our summarizing model proposed)
Our checking model objected to how our summarizing model worded a finding about this race. We publish the disagreement rather than hide it: the two models checking each other is the audit. The objection: Body claims 'All three advanced from the top four primary' but final_order for AK-SS-O lists only 3 candidates total (Rauscher, Sheldon, Bauer), contradicting the idea of a 'top four' from which three advanced out of four. This appears to be a factual inconsistency about the primary field size not supported by the provided data, and the piece mixes present-tense chatbot description with past-tense election result narration in a way that obscures whether this is live tracking or historical retrospective.
Written by GPT-5.5, disputed by Claude Sonnet 5