Midterm night: Cooper, Ossoff, Bottoms win as Supreme Court rejects DOJ ballot bid
In short
- Roy Cooper (D) wins North Carolina's open Senate seat over Michael Whatley on an AP call carried by 11 domains
- Jon Ossoff holds Georgia Senate vs. Rep. Mike Collins; Keisha Lance Bottoms wins the governor's race over Rick Jackson
- Supreme Court denies DOJ's emergency bid to set aside ~20,000 California ballots, acting within hours and without noted dissents
- Ashley Moody (R) wins Florida's special Senate election over Angie Nixon; Chris Pappas projected winner in New Hampshire
- CBS projects Rebecca Bennett over Rep. Tom Kean Jr. in NJ-7 and Stefany Shaheen in NH-1 amid ballot-scanner problems in Westchester
- 155 of 160 articles assessed real, 5 mixed, none flagged fabricated — but many verdicts rest on feed summaries alone
The full briefing
Election Day for the November 3, 2026 midterms dominated the cycle, and the results coverage was heavy on race calls. NPR reported, on an Associated Press call, that former Gov. Roy Cooper won North Carolina's open Senate seat over Republican Michael Whatley, the former RNC chair backed by President Trump to replace the retiring Thom Tillis. NPR also reported Jon Ossoff holding his Georgia seat against Rep. Mike Collins, and Keisha Lance Bottoms winning the Georgia governorship over Republican healthcare executive Rick Jackson — described as the first Black woman elected governor in the United States and the first Democrat to win that office in Georgia in more than two decades. NBC News projected Republican Sen. Ashley Moody over Democrat Angie Nixon in Florida's special election for Marco Rubio's former seat, and projected Democratic Rep. Chris Pappas the winner in New Hampshire. CBS News ran a House live blog projecting Rebecca Bennett over Rep. Tom Kean Jr. in New Jersey's 7th and Stefany Shaheen in New Hampshire's 1st, alongside exit-poll findings on Trump's approval, ballot-scanner problems in Westchester County, and quotes from Hakeem Jeffries and Mike Johnson.
The other major thread was legal. The Washington Examiner, the New York Times, and by corroboration NPR, AP and Al Jazeera reported that the Supreme Court on Monday, Nov. 2, declined the Trump administration's emergency application to force California election officials to set aside roughly 20,000 ballots tied to voters the Justice Department suspects are noncitizens. The justices acted within hours of the filing and without noted dissents, leaving lower-court rulings against the administration intact. Separately, Breitbart covered Trump's closing midterm message, quoting him calling the Democratic slate "the most extreme and dangerous slate" in American history and touting border enforcement, the One Big Beautiful Bill's tax provisions and oil production. That piece was assessed as genuine reporting of a real speech, with the explicit caveat that several statistics inside the quotes are contested and go unchallenged in the text.
The verification split was unusually clean: 155 of 160 articles assessed as accurate reporting, five as mixed, and nothing flagged as fabricated. That is less reassuring than it sounds. A large share of the "real" verdicts — the Nebraska live-results page, the Kentucky and Michigan results trackers, the Times midterm explainer, the Pappas video item — were judged from one-sentence RSS summaries with no body text, meaning the assessment rests on outlet reputation and on the fact that many newsrooms ran parallel election-night pages, not on any verified vote data. Several reviews say so plainly: corroboration here confirms that an election happened and that AP calls were made, not that any individual outlet's numbers are correct.
One pattern is worth flagging. The single substantive mixed verdict, on a TimesNow live blog, was low confidence (0.45) but for concrete reasons: the headline's claim that Trump approval is "near record low, say reports" appears nowhere in the body, and the page carries an unattributed ballot-interference allegation against Alaska Senate candidate Mary Peltola. That is distortion by aggregation rather than invention, and it is the kind of claim to watch for corroboration before repeating. Elsewhere, assessors noted details they could not check at all: the Guardian's election-day guide references a war with Iran, two citizens killed by federal immigration agents, Lindsey Graham's death in July, and a Justice Department effort to delay Alaska certification. An NPR photo gallery independently describes a Minneapolis memorial for Alex Pretti, said to have been killed with Renée Good by federal immigration officers in January — partial cross-corroboration of one of those strands, from a different outlet. A Technical.ly results hub also referenced a 2026 North Dakota Senate race that does not fit the standard Senate class rotation.
What to watch: House control was still being counted when these pieces filed, so the CBS projections are partial. Watch for the Alaska certification dispute, Louisiana's December runoffs under its open-primary system, and whether the Peltola allegation and the Guardian's unverified national-conditions claims pick up sourcing from independent newsrooms. And treat results-tracker pages as unverified until the full rendered numbers can be inspected.
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 →