Alaska spoiler scandal and 43 million early ballots dominate the final midterm week
In short
- DOJ subpoenas and a state criminal probe hit Mary Peltola's Alaska Senate bid over an alleged second "Dan Sullivan"
- AP: 43M+ early ballots cast five days out; Texas up ~50% at 6.3M, South Carolina up 83%, Maine/Michigan/NC/Kansas up 30%+
- Cook Political Report shifted 12 more House races toward Democrats Friday, now rating 218+ seats lean-D or better
- DOJ issued grand jury subpoenas to LA County election workers, appearances set Nov. 4, after Judge Carter rebuked the department
- Pope Leo XIV requested an Illinois absentee ballot Oct. 3; outlets report him as the first pope to vote in a US election
- Of 120 articles assessed, 1 was flagged fabricated and 5 mixed e none among the most widely circulated items
The full briefing
Five days out from the November 3 midterms, the Alaska Senate race took over the cycle. Multiple outlets reported that the Justice Department issued dozens of subpoenas tied to "ballot-stacking" allegations against Democratic nominee Mary Peltola, whose campaign is accused of recruiting a man sharing the name of Republican incumbent Sen. Dan Sullivan to appear on the ballot. NBC reported Alaska's governor ordered a criminal investigation; Breitbart reported Trump, at a South Carolina rally, called the effort a "criminal act" and said Peltola should be removed from the ballot and prosecuted. The Hill reported Peltola, speaking in Juneau, again denied knowledge of the scheme and said she wanted to "clear the air." All of this traces back to a New York Times investigation describing an internal "ballot composition project," quoting former Peltola campaign manager Elisa Rios saying the candidate was "definitely aware."
The other dominant thread was turnout and the map. An AP data analysis carried by the Chicago Tribune and others put early voting above 43 million ballots nationally, roughly 50 percent ahead of 2022 in Texas at 6.3 million, with gains over 30 percent in Maine, Michigan, North Carolina and Kansas and 83 percent in South Carolina. The Cook Political Report moved another 12 House races toward Democrats, including Republican-held seats such as Brian Fitzpatrick's PA-01 and Victoria Spartz's IN-05, and now rates more than 218 seats as lean-Democratic or better. Separately, NPR reported DOJ grand jury subpoenas to Los Angeles County Registrar-Recorder staff, with appearances set for Nov. 4, in a fight over federal demands to sequester ballots from roughly 20,000 people flagged as "likely" noncitizens e a request Judge David O. Carter had already rejected while admonishing the department. AP's Decision Notes flagged Minnesota's open governor and Senate races, and a wire item noted lingering Democratic divisions in Michigan's Senate contest. The lighter story of the day: Pope Leo XIV, a registered Illinois voter, requested an absentee ballot on Oct. 3 and is reported to have cast it, which several outlets called a first for any pope in a US election.
The verification picture was unusually clean. Of 120 articles assessed, 95 percent were marked real, one was flagged as fabricated or machine-generated, and five were assessed as mixed. None of the flagged items appeared among the most widely picked-up stories, meaning the high-volume coverage this cycle came from wire copy and established newsrooms rather than from the low-trust aggregation that usually spikes in a final week. That said, the confidence attached to these labels is modest: most of the top stories scored between 0.68 and 0.87, and the assessments repeatedly note that the 2026-specific facts e candidate matchups, poll margins, turnout figures e could not be independently checked.
Two patterns deserve attention. First, apparent corroboration is thinner than it looks. The Alaska cluster, the pope story, and the Cook ratings story were each carried by several domains, but in most cases those outlets are running the same AP or NYT reporting rather than confirming it independently; several verdicts were reached from RSS summaries alone, with the full article body never retrieved. Second, one premise is circulating without visible corroboration: the AP turnout analysis refers to a South Carolina special election following the death of Sen. Lindsey Graham, which the assessment explicitly could not confirm, even as CBS separately reported Trump campaigning in a surprisingly close South Carolina Senate race. Minor errors also surfaced in otherwise solid pieces e garbled station call signs in NPR's voter guide roundup, and the Haitian Times describing Mike Rogers as Michigan's Republican incumbent in an open-seat race.
Watch the Alaska investigation, where the gap between a Times investigation, a governor-ordered criminal probe and a presidential demand for prosecution is where distortion is most likely to creep in. Watch the Los Angeles subpoenas as the 9th Circuit appeal and the Nov. 4 appearance dates land just after polls close. And watch whether the early-voting numbers hold up against final turnout e outsized first-week figures are the kind of statistic that gets stripped of its comparison baseline as it travels.
Verdict split for this briefing
Every verdict behind this briefing is an automated assessment with a stated confidence, not a determination of fact. What we can and cannot measure →