Alaska U.S. House At-Large Race 2026

Ten days after Alaska voters picked Begich, chatbots still rate him far above his actual primary share

Alaska voters decided the U.S. House primary on August 18, ten days ago. Republican incumbent Begich finished first at 46.1 percent and advanced, along with Hill, an independent, at 30.9 percent, and Democrat Schultz at 8.0 percent. Democrat Williams did not advance, at 2.6 percent. The chatbots picked Begich first too, matching voters on the winner and the order all the way down, but they had rated him at 68.2 percent, 22.1 points above his real total. That gap has been the running story here for over a week and has not closed. The count is still unofficial, with 400 of 403 precincts reporting.

Alaska State Senate District O Race 2026

Ten days after Alaska voted, chatbots still rank Bauer above Sheldon in District O

All three District O candidates advance to November regardless, since this top four primary only had three names on the ballot. But the chatbots still have not caught up to what voters actually did on August 18. Republican incumbent Rauscher topped both the panel and the vote, chatbots picked him at 42.2 percent and voters gave him 39.8 percent. The miss is lower down. Chatbots put Democrat Peter Bauer second at 33.1 percent, ahead of Republican Ryan Sheldon at 15.3 percent. Voters said the opposite, Sheldon took 31.6 percent to Bauer's 28.6 percent. That is a 16.3 point gap between what the chatbots say about Sheldon and what he actually got. The count is unofficial, with 400 of 403 precincts in.

Alaska Governor Race 2026

The governor market picked Wilson while voters picked Kreiss-Tomkins, because Alaska apparently needed one more split screen.

In the governor primary, the unofficial count put Jonathan Kreiss-Tomkins first at 21.6 percent, while markets gave Bernadette Wilson the top win probability at 36.5 percent. The chatbot panel most often named Thomas Scott Begich at 18.9 percent, not Kreiss-Tomkins, who was at 11.1 percent.

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

“Ten days after Alaska voters picked Kreiss-Tomkins for governor, chatbots finally stopped naming Walker and picked the wrong winner instead” (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 calls the fourth place advancement safe even though the unofficial cutline margin is only 1.0 point.

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

“Ten days after Alaska voters picked Peltola for Senate, chatbots still favor Sullivan” (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 parentheses around R. It also frames a market probability as closer to the result in a race that already has an unofficial vote count.