https://civly.ai/newsroom Civly newsroom: press releases, media coverage, and insights from the team behind the AI campaign intelligence platform. en-us Wed, 26 Aug 2026 00:00:00 +0000 https://civly.ai/newsroom/lexis-cleanup https://civly.ai/newsroom/lexis-cleanup Wed, 26 Aug 2026 00:00:00 +0000 Matthew Hanauer Measured across 29 client LexisNexis name searches and the 33,771 articles they returned during the 2026 cycle. On the fourteen searches whose duplicate pass was logged article by article, 7,457 of 23,792 articles were copies of another article already in the same file — 31%, and as high as 46% on a single Iowa House race, where 245 original stories accounted for 985 articles. On the eleven searches carrying a categorized exclusion log that names the person behind every removal, 3,527 of 19,292 articles were about somebody else or said nothing about the subject at all — 18%, but ranging from 1% to 94% depending on how common the candidate's name is. Taken together on those eleven searches, 9,435 of 19,292 articles were removed as either a duplicate or not-this-person: 49%. In the worst case, a Florida House race whose candidate's surname is also an ordinary English word for a job, 2,997 articles returned 115 that reached the research file. Removed articles are preserved and enumerated with the reason for each removal rather than deleted. https://civly.ai/newsroom/progressive-primary-challenges https://civly.ai/newsroom/progressive-primary-challenges Fri, 21 Aug 2026 00:00:00 +0000 Matthew Hanauer An analysis of the 45 Democratic primaries from 2016 to 2026 in which a sitting U.S. House member faced a challenger scored to their left, built from 6,948 OpenElections result files plus published election boxes, FEC bulk filings, and MIT Election Lab general-election results. Nine challenges won, about one in five against roughly one in twenty-five for challenges generally. Money did not separate the two groups: the challenger was outspent by roughly 2.5 to 1 whether the challenge won or lost, and the incumbents who were beaten had raised four times what an incumbent who won their primary raised. What separated them was the number of donors — a median 905 against 258, a 3.5-fold gap that holds at 2.2-fold inside safe Democratic seats, 2.5-fold per 1,000 general-election voters, and 1,318 against 174 inside New York alone, where every candidate draws on one donor pool. Eight of the nine wins came in a midterm year, in seats that went 78.0% Democratic at the last general against 63.9% for the challenges that lost. Of the six contested primaries still to vote in 2026, only Massachusetts’s 8th — Patrick Roath against Stephen Lynch — has the profile. Stated limits: the finding is descriptive, not causal, because a challenger who was always going to win also attracts more donors; nine wins is a small base; half the primary dates are estimated; and the 2026 cycle is unfinished. https://civly.ai/newsroom/ai-at-the-ballot-box https://civly.ai/newsroom/ai-at-the-ballot-box Thu, 20 Aug 2026 00:00:00 +0000 Zack Czajkowski A projection of how many American voters will use AI on their November 2026 ballot, built by re-weighting Pew’s February 2026 chatbot-use crosstabs from adults onto the midterm electorate and applying the conversion rates measured after Scotland’s May 2026 election: about 16 million will use AI to help decide a contest, 42 million will run an election-related query, and 69 million will read AI-generated text while researching their ballot. Documents the four observed voter behaviors from published reporting — photographing the ballot, the refusal-and-reframe loop, values-matching against a long ballot, and unrequested AI search summaries — alongside audit results on accuracy and completeness, the campaign-side monitoring industry, and falsifiable predictions for any pollster who asks before November. https://civly.ai/newsroom/donor-breadth-vote-share https://civly.ai/newsroom/donor-breadth-vote-share Sun, 16 Aug 2026 00:00:00 +0000 Matthew Hanauer A model of contested U.S. House and Senate races from 1980 to 2024, built from MIT Election Lab returns and FEC bulk filings, measuring what each campaign fundraising decision is worth in points of vote share. Doubling the number of separate organizations giving, holding total dollars fixed, is worth +1.11 points to a challenger and -0.48 to an incumbent — opposite signs that cancel in any pooled average. Doubling the money itself is the only lever with the same sign for everyone, and its return is largest for the smallest campaigns. Trained on 1980-2014 and scored on the 2016 and 2018 elections it had never seen, the model misses vote share by 3.69 points against 5.69 for knowing only the seat and the incumbent. https://civly.ai/newsroom/red-flags-were-already-public https://civly.ai/newsroom/red-flags-were-already-public Fri, 14 Aug 2026 00:00:00 +0000 Civly The standard defence of a bad signing is that nobody could have known. In thirteen documented cases across the NFL, MLB and world football, the warning was already in a court file, a published sanction or a video that had been on the news, sometimes for years. Each entry records what was available, how long before the signature, and what the club paid. https://civly.ai/newsroom/union-political-money https://civly.ai/newsroom/union-political-money Thu, 13 Aug 2026 00:00:00 +0000 Matthew Hanauer A test of the five separate claims behind union political spending, each against a control group of unfunded legislators in the same party: $175.4M of union ballot-measure money across 12 states, 5,361 candidates in six states with certified results, and 55 within-party legislator comparisons drawn from LM-2 filings, state campaign finance, and state legislative records. https://civly.ai/newsroom/uncontested-courthouse-races https://civly.ai/newsroom/uncontested-courthouse-races Thu, 13 Aug 2026 00:00:00 +0000 Matthew Hanauer A national count of uncontested elections for prosecutor, judge and sheriff, built from the MIT Election Data and Science Lab's certified precinct returns for 2022 and 2024: 6,358 races across 45 states, of which 4,240 had no opponent on the ballot. Includes the party split by office — wide for sheriffs, absent for judges — and the size distribution showing a median uncontested race of 16,520 votes. https://civly.ai/newsroom/nil-followers-and-pay https://civly.ai/newsroom/nil-followers-and-pay Tue, 04 Aug 2026 00:00:00 +0000 Matthew Hanauer A study of 112 college athletes with published contract values between $500,000 and $6.5M, testing whether social audience predicts roster pay once position, on-field production and program revenue are controlled. Production and program scale move pay; follower count and engagement rate cannot be distinguished from no effect. https://civly.ai/newsroom/poll-of-ais https://civly.ai/newsroom/poll-of-ais Fri, 17 Jul 2026 00:00:00 +0000 Matthew Hanauer Five AI models ran the identical synthetic poll on the identical simulated voters across eight certified elections. The panel average beat every single model, and disagreement between models flagged the one race they all misread. https://civly.ai/newsroom/maine-senate-2026 https://civly.ai/newsroom/maine-senate-2026 Sat, 11 Jul 2026 00:00:00 +0000 Matthew Hanauer A synthetic poll of Maine's 2026 Senate race: nine potential Democratic nominees tested head-to-head against Susan Collins on a simulated panel of 4,597 real registered Maine voters, validated against five certified Maine elections. https://civly.ai/newsroom/turnout-score https://civly.ai/newsroom/turnout-score Mon, 06 Jul 2026 00:00:00 +0000 Matthew Hanauer A modeled turnout-propensity score graded against real recorded turnout across nine states: 84% of voter-level turnout calls correct versus 78% for the standard participation-rate rule, and 83% versus 67% on a low-turnout municipal race. https://civly.ai/newsroom/primary-scorecard https://civly.ai/newsroom/primary-scorecard Fri, 26 Jun 2026 00:00:00 +0000 Matthew Hanauer An AI voter-simulation forecast of every contested New York and Maryland House primary — published in advance and graded against every certified result. https://civly.ai/newsroom/press-release-ai-stance https://civly.ai/newsroom/press-release-ai-stance Fri, 01 May 2026 00:00:00 +0000 Civly Civly's stance on artificial intelligence, and why the party of science can't afford to look away. https://civly.ai/newsroom/press-release-launch https://civly.ai/newsroom/press-release-launch Thu, 01 Jan 2026 00:00:00 +0000 Civly Civly launches its AI-powered campaign intelligence platform for political campaigns.