Tech Policy Press is a nonprofit media and community venture intended to provoke new ideas, debate and discussion at the intersection of technology and democracy. We publish opinion and analysis.
Tech Policy Press - Technology and Democracy
Tech Policy Press is a nonprofit media and community venture intended to provoke new ideas, debate and discussion at the intersection of technology and democracy. We publish opinion and analysis.
State media control influences large language models
Millions of people around the world query large language models (LLMs) for information. Although several studies have compellingly documented the persuasive potential of these models1–10, there is limited evidence of who or what influences the models themselves, leading to a flurry of concerns about which companies and governments build and regulate the models. Here we show through six studies that government control of the media across the world already influences the output of LLMs via their training data. We use a cross-national audit to show that LLMs exhibit a stronger pro-government valence in the languages of countries with lower media freedom than in those with higher media freedom. This result is correlational, so to triangulate the specific mechanism of how state media control can influence LLMs, we develop a multi-part case study on China’s media. We demonstrate that media scripted and curated by the Chinese state appears in LLM training datasets. To evaluate the plausible effect of this inclusion, we use an open-weight model to show that additional pretraining on Chinese state-coordinated media generates more positive answers to prompts about Chinese political institutions and leaders. We link this phenomenon to commercial models through two audit studies demonstrating that prompting models in Chinese generates more positive responses about China’s institutions and leaders than do the same queries in English. The combination of influence and persuasive potential across languages suggests the troubling conclusion that states and powerful institutions have increased strategic incentives to leverage media control in the hopes of shaping LLM output.
State media control influences large language models
Millions of people around the world query large language models (LLMs) for information. Although several studies have compellingly documented the persuasive potential of these models1–10, there is limited evidence of who or what influences the models themselves, leading to a flurry of concerns about which companies and governments build and regulate the models. Here we show through six studies that government control of the media across the world already influences the output of LLMs via their training data. We use a cross-national audit to show that LLMs exhibit a stronger pro-government valence in the languages of countries with lower media freedom than in those with higher media freedom. This result is correlational, so to triangulate the specific mechanism of how state media control can influence LLMs, we develop a multi-part case study on China’s media. We demonstrate that media scripted and curated by the Chinese state appears in LLM training datasets. To evaluate the plausible effect of this inclusion, we use an open-weight model to show that additional pretraining on Chinese state-coordinated media generates more positive answers to prompts about Chinese political institutions and leaders. We link this phenomenon to commercial models through two audit studies demonstrating that prompting models in Chinese generates more positive responses about China’s institutions and leaders than do the same queries in English. The combination of influence and persuasive potential across languages suggests the troubling conclusion that states and powerful institutions have increased strategic incentives to leverage media control in the hopes of shaping LLM output.
PropagandaScope
Tracking articles from newspapers. Real-time analysis of propaganda transmission patterns.
Manufacturing Consent: The Political Economy of the Mass Media
A bold and eye-opening exposé on how power and propaganda distort the news, now more relevant than ever • With an updated introduction“[A] compelling indictment of the news media’s role in covering up errors and deceptions in American foreign policy.”—The New York Times Book Review Renowned scholars Edward S. Herman and Noam Chomsky reveal how U.S. news media, far from being independent watchdogs, often function as tools of elite influence. With probing analysis, they present their Propaganda Model, a framework that explains how systemic bias shapes the stories we’re told, the voices we hear, and the truths that remain hidden.Through deeply researched case studies, from the Vietnam War to coverage of “worthy” vs. “unworthy” victims, Manufacturing Consent exposes the structural forces that drive news organizations to reinforce power rather than question it. It’s a sobering portrait of a media system more interested in maintaining order than informing the public.This edition includes an introduction updating key examples and expanding the Propaganda Model’s relevance to issues like the coverage of NAFTA, the media’s treatment of global protests, and environmental regulation.Manufacturing Consent is a powerful assessment of how propagandistic the U.S. mass media are, how they systematically fail to live up to their self-image as providers of the kind of information that people need to make sense of the world, and how we can understand their function in a radically new way.Whether you’re a student, activist, or citizen looking to see beyond the headlines, this book will transform how you understand the media—and the world around you.
How propaganda exploits the infrastructure of truth: A case study of #IStandWithPutin
This paper investigates the #IStandWithPutin hashtag campaign as a case study to conceptualise “infrastructural propaganda,” an influence strategy that manipulates the invisible processes and syste...
How propaganda exploits the infrastructure of truth: A case study of #IStandWithPutin
This paper investigates the #IStandWithPutin hashtag campaign as a case study to conceptualise “infrastructural propaganda,” an influence strategy that manipulates the invisible processes and syste...
The Rise of the Right-Wing Media Machine
This conservative media machine now rivals—and may well surpass—the power and the influence of the old-line press. It holds sway over much of the national agenda, deciding which ideas and...
Replication Data for "State Media Control Influences Large Language Models"
Replication dataset for "State Media Control Influences Large Language Models," forthcoming in Nature (https://doi.org/10.1038/s41586-026-10506-7). We show through six studies that government control of the media across the world influences the output of large language models (LLMs) via their training data.
Replication Data for "State Media Control Influences Large Language Models"
Replication dataset for "State Media Control Influences Large Language Models," forthcoming in Nature (https://doi.org/10.1038/s41586-026-10506-7). We show through six studies that government control of the media across the world influences the output of large language models (LLMs) via their training data.
Show your work: The new terms for trust in journalism - PressThink
The transparency movement has finally come of age. Power has shifted to the users. Their trust has to be earned in different ways now.
AI Policy • States Newsroom
AI Policy States Newsroom is, above all, dedicated to journalism practiced by people. We believe in the power of having on-the-ground reporters covering our communities with the finesse and nuance that only human interaction can produce. We also recognize the power and potential of generative AI. We are open to the responsible and transparent use […]
Why Don’t We Learn from Social Media? Studying Effects of and Mechanisms behind Social Media News Use on General Surveillance Political Knowledge
Does exposure to news affect what people know about politics? This old question attracted new scholarly interest as the political information environment is changing rapidly. In particular, since c...
Artificial Intelligence in the News: How AI Retools, Rationalizes, and Reshapes Journalism and the Public Arena
<p>Download the pdf here. Executive Summary Despite growing interest, the effects of AI on the news industry and our information environment — the public arena — remain poorly understood. Insufficient attention has also been paid to the implications of the news industry’s dependence on technology companies for AI. Drawing on 134 interviews with news workers […]</p>
Push journalism vs. pull journalism
Let's experiment with making news products so relevant that people pull them into their lives
The Metric Media network runs more than 1,200 local news sites. Here are some of the non-profits funding them.
<p>This is the first part of a two-part investigation. The second piece can be found here. As local news outlets across the US struggle to stay afloat, pop-up newsrooms are filling the void. Organizations with ties to political parties, special-interest groups, and lobbyists have emerged under the guise of news. One of the biggest such […]</p>
