State media control influences large language models | Nature
Subjects
- Computer science
- Politics
- Society
Abstract
Millions of people around the world query large language models (LLMs) for information. Although several studies have compellingly documented the persuasive potential of these models1,2,3,4,5,6,7,8,9,10, 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.
Access through your institution
Buy or subscribe
This is a preview of subscription content, access via your institution
Access options
Access through your institution
Access Nature and 54 other Nature Portfolio journals
Get Nature+, our best-value online-access subscription
$32.99 / 30 days
cancel any time
Learn more
Subscribe to this journal
Receive 51 print issues and online access
$199.00 per year
only $3.90 per issue
Learn more
Buy this article
- Purchase on SpringerLink
- Instant access to the full article PDF.
USD 39.95
Prices may be subject to local taxes which are calculated during checkout
Fig. 1: Logical flow of the six studies.The alternative text for this image may have been generated using AI.
Fig. 2: Chinese state-coordinated media is in the training data of commercial language models.The alternative text for this image may have been generated using AI.
Fig. 3: Additional pretraining on state-coordinated media causes pro-Chinese government slant (study 3).The alternative text for this image may have been generated using AI.
Fig. 4: Commercial models give responses more favourable to China’s political institutions when prompted in Chinese.The alternative text for this image may have been generated using AI.
Fig. 5: Language-exclusive countries are rated more favourably in their own language when they have lower media freedom (study 6).The alternative text for this image may have been generated using AI.
Data availability
Derivative data products are available in our replication archive (https://doi.org/10.7910/DVN/NECR2K). We released transformed products only rather than the full text of raw news stories because we do not hold their copyright. Our full-text articles were collected through a combination of news website scraping and data purchases from WisersOne (formerly WiseNews). We have provided additional replications of the studies using the latest models at the time of publication (https://state-media-influence-llm.github.io/).
Code availability
The replication code for all analyses in the main text and extended data is available in our replication archive (https://doi.org/10.7910/DVN/NECR2K).
References
- Palmer, A. & Spirling, A. Large language models can argue in convincing ways about politics, but humans dislike AI authors: implications for governance. Polit. Sci. 75, 281–291 (2023).
Article
Google Scholar
- Bai, H. et al. LLM-generated messages can persuade humans on policy issues. Nat. Commun. 16, 6037 (2025).
Article
ADS
PubMed
PubMed Central
Google Scholar
- Hackenburg, K. & Margetts, H. Evaluating the persuasive influence of political microtargeting with large language models. Proc. Natl Acad. Sci. USA 121, e2403116121 (2024).
Article
PubMed
PubMed Central
Google Scholar
- Salvi, F. et al. On the conversational persuasiveness of GPT-4. Nat. Hum. Behav. 9, 1645–1653 (2025).
Article
PubMed
PubMed Central
Google Scholar
- Costello, T. H., Pennycook, G. & Rand, D. G. Durably reducing conspiracy beliefs through dialogues with AI. Science 385, eadq1814 (2024).
Article
ADS
CAS
PubMed
Google Scholar
- Carrasco-Farre, C. Large language models are as persuasive as humans, but how? About the cognitive eff