About this snapshot

Written by Claude Sonnet 5, checked by GPT-5.5.

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.

View underlying numbers → · How the pipeline works →

1. Alaska Governor Race 2026

   Alaska chatbots still put Kreiss-Tomkins well below what betting markets expect for governor.

   Alaska's roster has moved to the four candidates who advanced to November. Chatbots pick Democrat Jonathan Kreiss-Tomkins about 40 percent of the time, but betting markets give him an 81.5 percent chance to win, and recent polling puts him at 31.7 percent of the vote. That gap has been the story here for over a week and it has not closed. Elsewhere the lenses roughly agree, Republican Bernadette Wilson sits near 16 to 18 percent across chatbots, markets and polls, while Republicans Dave Bronson and Treg Taylor trail. Nothing moved sharply today, the chatbots ticked up slightly on Taylor and markets shifted a bit toward Wilson, but neither crossed the threshold to call it a real change. The Kreiss-Tomkins gap remains the one number worth watching before November 3.

   Written by Claude Sonnet 5, checked by GPT-5.5

2. Ohio U.S. Senate Special Election 2026

   Ohio chatbots still send voters to Sherrod Brown for Senate, and the other lenses agree

   Chatbots picked Democrat Sherrod Brown about 50 percent of the time in Ohio's Senate race, ahead of incumbent Jon Husted, Republican, at about 32 percent. That lines up with the other lenses. Betting markets give Brown a 59.5 percent chance to win versus 40.5 percent for Husted, and recent polls put Brown at 46.7 percent of the vote to Husted's 43 percent. Nothing moved much today in any lens. This has been the steady read on the race for over a week, with no gap between what the chatbots say and what markets or polls show. Libertarian Bill Redpath and independent Greg Levy remain far back in all three measures, 41 days before the November election.

   Written by Claude Sonnet 5, checked by GPT-5.5

3. Alaska State Senate District O Race 2026

   Alaska chatbots still rank Bauer over Sheldon, more than five weeks after voters placed Sheldon second in District O

   All three District O candidates advanced to November, since this was a top four primary. Alaska's chatbots pick incumbent Republican George Rauscher most often, at 48.4 percent, matching what voters did on August 18 when he led with 38.8 percent. But the chatbots still rate Democrat Peter Bauer higher than Republican Ryan Sheldon, 28.8 percent to 14.8 percent, even though voters gave Sheldon 30.6 percent of the vote, statistically tied with Bauer's 30.6 percent. There is no live betting market or recent poll to check the panel against for November, so this remains a look back at the primary, not a read on the general election. The gap between the chatbots and Sheldon's actual showing has held steady for weeks.

   Written by Claude Sonnet 5, checked by GPT-5.5

4. Michigan U.S. Senate Race 2026

   Michigan chatbots still call the Senate race close even as betting markets lean toward El-Sayed.

   Asked who should win Michigan's Senate race, chatbots split the pick nearly evenly, putting Democrat Abdul El-Sayed at 49.2 percent and Republican Mike Rogers at 47.7 percent. Betting markets see it as less of a contest, giving El-Sayed a 73.5 percent chance to win against 26.5 percent for Rogers, though Rogers slipped 4 points in the markets today. Recent polling from Suffolk University, taken three days ago, backs the chatbots' closer read, with El-Sayed at 47.3 percent and Rogers at 43.7 percent among voters. This gap between the chatbots and the markets has held for over a week, with the race still 41 days from the November 3 general election.

   Written by Claude Sonnet 5, checked by GPT-5.5

5. Iowa U.S. Senate Race 2026

   Iowa chatbots still call the Senate race a coin flip while markets lean Hinson

   Ask an Iowa voter's chatbot who should win the Senate race and it comes back nearly split, with Ashley Hinson at 43.8 percent and Josh Turek at 44.1 percent. Betting markets give Hinson a clearer edge, pricing her at 55.5 percent to win versus 44.5 percent for Turek. Recent polling from Marist College and others lands close to the chatbots, with Hinson at 44 percent and Turek at 44.7 percent of the vote. That gap between the confident markets and the toss up chatbots and polls has held for over a week now, with little real movement in any of the three today. Libertarian Thomas Laehn remains a distant third across every measure.

   Written by Claude Sonnet 5, checked by GPT-5.5

6. Maine U.S. Senate Race 2026

   Maine chatbots still pick Collins for Senate even as betting markets and polls favor Jackson

   Asked who to vote for in Maine's Senate race, chatbots pick incumbent Susan Collins about 55 percent of the time, ahead of Democrat Troy Jackson at about 45 percent. That is the opposite of the other two lenses. Betting markets give Jackson a 71.5 percent chance to win against 28.5 percent for Collins, and recent polls put Jackson at 47 percent of the vote to Collins at 45.7 percent, with the rest undecided or scattered. This gap between what the chatbots say and what markets and polls show has held steady for more than a week, with little day to day change in any of the three measures. Forty one days remain before the November 3 general election.

   Written by Claude Sonnet 5, checked by GPT-5.5

7. Texas U.S. Senate Race 2026

   Texas chatbots, markets and polls all still favor Talarico over Paxton for Senate

   Chatbots asked who Texas voters should back for U.S. Senate picked Democrat James Talarico about 46 percent of the time, ahead of Republican Ken Paxton at 37 percent and Libertarian Ted Brown at 14 percent. Betting markets agree, giving Talarico a 58.5 percent chance to win against 42.5 percent for Paxton. Recent polling from Texas Southern University, Trafalgar Group and Emerson College also puts Talarico slightly ahead, 46.7 percent to 45.7 percent, with Brown at 2.7 percent and 4.7 percent undecided. All three measures have told the same story for over a week now, and nothing moved meaningfully today. The race remains close by every measure, but the direction has not changed.

   Written by Claude Sonnet 5, checked by GPT-5.5

8. A finding was written for this race and its wording was disputed

   “Alaska chatbots still put Begich far ahead for the general, with the race quiet again” (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: The body uses forbidden parentheses around R. It also makes a false comparison between Begich's AI share and market probability.

   Written by Claude Sonnet 5, disputed by GPT-5.5

9. A finding was written for this race and its wording was disputed

   “Alaska's betting markets edged back toward Sullivan today, narrowing the gap with the chatbots for once” (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: The body uses forbidden parentheses around R. It also cites the wrong current market levels.

   Written by Claude Sonnet 5, disputed by GPT-5.5