The 2026 midterms: why we should be wary of what GenAI tells us ahead of next year’s elections

Peter Finn; Lauren C. Bell; Amy Tatum; Caroline V. Leicht

The 2026 midterms: why we should be wary of what GenAI tells us ahead of next year’s elections

Peter Finn
, Lauren C. Bell
, Amy Tatum
, Caroline V. Leicht

University of Southampton
Randolph-Macon College
Bournemouth University

Research output: Contribution to specialist publication › Article

Abstract

Generative Artificial Intelligence (GenAI) is becoming increasingly part of the lives of many Americans, including seeking information about politics and elections. Less than one year out from the 2026 midterms, Peter Finn, Lauren C. Bell, Amy Tatum, and Caroline V. Leicht look at the effectiveness of GenAI as a tool for searching for information relevant to elections. They find that GenAI tends to skew its answers to the national level and can hallucinate answers to questions about politics which require nuanced answers.

This article is part of our ‘The 2026 Midterms‘ series curated by Peter Finn (Kingston University). The series will explore the run-up to the 2026 US midterm elections at the state and national levels.

Original language	English
Specialist publication	LSE USAPP United States Politics and Policy Blog
Publisher	London School of Economics
Publication status	Published - 11 Dec 2025

Bibliographical note

This article is based on the paper, “Generative AI and the Nationalization of US Politics” in the Journal of American Studies.

Keywords

GenAI
midterms

Access to Document

https://blogs.lse.ac.uk/usappblog/2025/12/11/the-2026-midterms-why-we-should-be-wary-of-what-genai-tells-us-ahead-of-next-years-elections/Licence: CC BY-NC

1 Article

Generative AI and the nationalisation of U.S. politics
Bell, L. C., Finn, P., Tatum, A. & Leicht, C. V., 9 Oct 2025, (E-pub ahead of print) In: Journal of American Studies.
Research output: Contribution to journal › Article › peer-review

Open Access
File
53 Downloads (Pure)

Cite this

@misc{2f76695581d9419bb5b8f3933efcb1a5,

title = "The 2026 midterms: why we should be wary of what GenAI tells us ahead of next year{\textquoteright}s elections",

abstract = "Generative Artificial Intelligence (GenAI) is becoming increasingly part of the lives of many Americans, including seeking information about politics and elections. Less than one year out from the 2026 midterms, Peter Finn, Lauren C. Bell, Amy Tatum, and Caroline V. Leicht look at the effectiveness of GenAI as a tool for searching for information relevant to elections. They find that GenAI tends to skew its answers to the national level and can hallucinate answers to questions about politics which require nuanced answers. This article is part of our {\textquoteleft}The 2026 Midterms{\textquoteleft} series curated by Peter Finn (Kingston University). The series will explore the run-up to the 2026 US midterm elections at the state and national levels. ",

keywords = "GenAI, midterms",

author = "Peter Finn and Bell, \{Lauren C.\} and Amy Tatum and Leicht, \{Caroline V.\}",

note = "This article is based on the paper, “Generative AI and the Nationalization of US Politics” in the Journal of American Studies.",

year = "2025",

month = dec,

day = "11",

language = "English",

journal = "LSE USAPP United States Politics and Policy Blog",

publisher = "London School of Economics",

}

TY - GEN

T1 - The 2026 midterms

T2 - why we should be wary of what GenAI tells us ahead of next year’s elections

AU - Finn, Peter

AU - Bell, Lauren C.

AU - Tatum, Amy

AU - Leicht, Caroline V.

N1 - This article is based on the paper, “Generative AI and the Nationalization of US Politics” in the Journal of American Studies.

PY - 2025/12/11

Y1 - 2025/12/11

N2 - Generative Artificial Intelligence (GenAI) is becoming increasingly part of the lives of many Americans, including seeking information about politics and elections. Less than one year out from the 2026 midterms, Peter Finn, Lauren C. Bell, Amy Tatum, and Caroline V. Leicht look at the effectiveness of GenAI as a tool for searching for information relevant to elections. They find that GenAI tends to skew its answers to the national level and can hallucinate answers to questions about politics which require nuanced answers. This article is part of our ‘The 2026 Midterms‘ series curated by Peter Finn (Kingston University). The series will explore the run-up to the 2026 US midterm elections at the state and national levels.

AB - Generative Artificial Intelligence (GenAI) is becoming increasingly part of the lives of many Americans, including seeking information about politics and elections. Less than one year out from the 2026 midterms, Peter Finn, Lauren C. Bell, Amy Tatum, and Caroline V. Leicht look at the effectiveness of GenAI as a tool for searching for information relevant to elections. They find that GenAI tends to skew its answers to the national level and can hallucinate answers to questions about politics which require nuanced answers. This article is part of our ‘The 2026 Midterms‘ series curated by Peter Finn (Kingston University). The series will explore the run-up to the 2026 US midterm elections at the state and national levels.

KW - GenAI

KW - midterms

M3 - Article

JO - LSE USAPP United States Politics and Policy Blog

JF - LSE USAPP United States Politics and Policy Blog

PB - London School of Economics

ER -

The 2026 midterms: why we should be wary of what GenAI tells us ahead of next year’s elections

Abstract

Bibliographical note

Keywords

Access to Document

Fingerprint

Research output

Generative AI and the nationalisation of U.S. politics

Cite this