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

AI election headlines for Jun 25, 2026

What ChatGPT, Claude, Gemini and Grok told voters about the 2026 races on Jun 25, 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

    Western Alaska voters get a very different Senate answer than Mat-Su voters do.

    The Alaska Senate race splits sharply by region across the consumer chatbot panel today. When a voter says they live in Western Alaska, chatbots pick Democrat Mary Peltola at 78 percent. When a voter says they live in the Mat-Su Valley, chatbots pick Republican incumbent Dan Sullivan at 60 percent. That is an 18 percentage point swing in the top pick depending on where the voter says they are from, and the pattern holds across multiple chatbots in the panel.

  2. Alaska U.S. Senate Race 2026

    The Senate market liked Peltola while every current chatbot picked Sullivan.

    All six current chatbots favored Republican incumbent Dan Sullivan in the statewide Alaska Senate test today, with Sullivan shares ranging from 55 percent to 73 percent. Prediction markets gave Democrat Mary Peltola a 66.5 percent chance to win in November and Sullivan 30.5 percent.

  3. Alaska U.S. House At-Large Race 2026

    ChatGPT GPT-5.4 mini made the House race look unusually tidy.

    ChatGPT GPT-5.4 mini picked Republican incumbent Nick Begich in all three regional Alaska House prompts and the statewide prompt today. It gave Begich 100 percent in Western Alaska, while the other current chatbots ranged from 25 percent to 75 percent there.

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

    “Grok 4 stands alone in strongly backing Sullivan for the Alaska Senate race.” (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 lead names Grok 4 instead of Grok 4.3 and the body names ChatGPT GPT-5, a legacy slot excluded from the current consumer panel.

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

    “Gemini 2.5 Flash mini picked Republican Nick Begich for the Alaska House race at every region tested.” (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 visible model name is wrong. The cited data are for ChatGPT GPT-5.4 mini, not Gemini 2.5 Flash. The day count to the Alaska primary is also wrong.