Skip to main content
Voting Monitor2026 Senate toss-ups and Alaska races
← All daily findings

AI election headlines for Jul 17, 2026

What ChatGPT, Claude, Gemini and Grok told voters about the 2026 races on Jul 17, 2026, set against prediction markets and recent polls. Each finding was written by one model and checked by another.

  1. Alaska U.S. Senate Race 2026

    Alaska's Senate race is still split three ways, with chatbots leaning Sullivan while markets and polls lean Peltola.

    Voters asking chatbots who should win Alaska's U.S. Senate race will most often get incumbent Republican Dan Sullivan, picked about 51.0 percent of the time by the AI panel. But the betting markets give Democrat Mary Peltola a 59.5 percent chance to win, and the polling average from NYT Siena and Alaska Survey Research puts her ahead too, at 47.5 percent of the vote to Sullivan's 45.1 percent. That gap between what the chatbots say and what the markets and polls show has held for more than a week, with no significant move in any lens today, 32 days before Alaska's top 4 primary.

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

    “Alaska chatbots still lean toward Walker for governor, even as markets favor Begich and Wilson” (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 contains forbidden colons. The body also exceeds the 120 word hard cap.

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

    “Chatbots and Alaska voters agree: incumbent Begich is well ahead in the U.S. House race, though the AI panel underrates Schultz next to the polls” (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 title and body contain forbidden colons. The body contains forbidden parentheses and a forbidden semicolon. The body exceeds the 120 word hard cap. The explanation that the AI panel has less written about Schultz to draw on is unsupported analysis.

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

    “Alaska's Senate race is still split three ways: chatbots lean Sullivan, markets and polls lean Peltola.” (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 title contains a forbidden colon. The body contains forbidden parentheses and a slash between words in NYT/Siena.

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

    “Chatbots still lean toward incumbent Rauscher in State Senate District O, with no market or poll data to check that against.” (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 contains forbidden parentheses and forbidden colons. The body exceeds the 120 word hard cap. The claim that voters have only that signal to go on overstates the dashboard signal.