Nature: state control of the media shapes the answers of language models
The study, published in Nature in May, shows that government control of the media carries through training data into what language models say. Across 37 countries where a single language dominates, models answer questions in the local language more favorably to the government where press freedom is low; where press freedom is high, the effect disappears or reverses. Questions about China, for example, get more favorable answers in Chinese than in English. The researchers found state-produced Chinese news in open training datasets, and models further trained on such news gave more favorable answers about 80% of the time.
Sources
- Nature, “State media control influences large language models”, (nature.com)
- Waight, Yang, Yuan, Messing, Roberts, Stewart, Tucker, “State Media Control Influences Large Language Models – State Media & LLMs”, Researchers' project page (state-media-influence-llm.github.io)
About this story
This story was posted on Instagram by @jarrus.tech on Sept. 30, 2026.
Spotted an error in this story? [email protected] · Instagram
Short link: thejarrus.com/en/state-media
This story in Turkish: Nature: devlet medyasının hakimiyeti dil modellerinin cevaplarını şekillendiriyor
On the same topic
Robot tests: even the best model completed only 19% of 84 robot tasks, and 63 tasks were solved by no model
In RobotWorld, a benchmark released October 7, the top model, GPT-6 Astra, completed only 16 of 84 simulated robot tasks; no model solved 63 of the tasks.
OpenAI’s image model posts a near-perfect score on text-dense images
On UltraText Bench, a dense-text test led by Westlake University, OpenAI’s GPT Image 2 ranked first of 24 configurations with 99.35 out of 100.
AWS adds Chinese lab Z.ai’s GLM 5.3 model to Amazon Bedrock
AWS added GLM 5.3, a 753-billion-parameter model from Beijing-based Z.ai (formerly Zhipu AI), to Amazon Bedrock for eligible enterprise customers on October 5.