Large language models (LLMs) are increasingly used in everyday tools and applications, raising concerns about their potential influence on political views. While prior research has shown that LLMs often exhibit measurable political biases--frequently skewing toward liberal or progressive positions--key gaps remain. Most existing studies evaluate only a narrow set of models and languages, leaving open questions about the generalizability of political biases across architectures, scales, and multilingual settings. Moreover, few works examine whether these biases can be actively controlled. In this work, we address these gaps through a large-scale study of political orientation in modern open-source instruction-tuned LLMs. We evaluate seven models, including LLaMA-3.1, Qwen-3, and Aya-Expanse, across 14 languages using the Political Compass Test with 11 semantically equivalent paraphrases per statement to ensure robust measurement. Our results reveal that larger models consistently shift toward libertarian-left positions, with significant variations across languages and model families. To test the manipulability of political stances, we utilize a simple center-of-mass activation intervention technique and show that it reliably steers model responses toward alternative ideological positions across multiple languages. Our code is publicly available at https://github.com/d-gurgurov/Political-Ideologies-LLMs.

While Large Language Models (LLMs) have made agentic AI, chatbots, and other intelligent applications possible, they have also enabled the affordable creation of highly convincing AI-generated disinformation (Bontcheva et al., 2024), which poses a systemic risk to democratic stability and global security (VIGINUM, 2025; Bengio, 2025). Initially, AI-generated texts suffered from linguistic mistakes and thus were more easily detectable by humans. However, modern LLMs, particularly instruction-tuned models, have significantly improved in producing outputs which are indistinguishable from human-written text (Spitale et al., 2023; Heppell et al., 2024). These advances have resulted in their misuse in generating persuasive disinformation narratives, including political manipulation, health disinformation, conspiracy propagation, and Foreign Information Manipulation and Interference (FIMI) (Vykopal et al., 2024; Chen and Shu, 2024a; Barman et al., 2024; Chen and Shu, 2024b; Heppell et al., 2024; VIGINUM, 2025). While there is a growing body of research on the generation and detection of LLM-produced disinformation (Chen and Shu, 2024a; Lucas et al., 2023; Vykopal et al., 2024; Heppell et al., 2024), a critical aspect remains largely unstudied - namely, whether LLMs are capable of generating fluent and convincing personalised disinformation (i.e., disinformation narratives tailored to specific audiences) in multiple languages and at scale. The few prior studies on AIgenerated personalised disinformation are limited to English and address a very narrow set of personas (e.g., students, parents) (Zugecova et al., 2024). Crucially, prior work has not yet examined whether LLMs can adapt disinformation to country-specific linguistic and cultural contexts in multiple languages.

We surveyed 582 AI researchers who have published in leading AI venues and 838 nationally representative US participants about their views on the potential development of AI systems with subjective experience and how such systems should be treated and governed. When asked to estimate the chances that such systems will exist on specific dates, the median responses were 1% (AI researchers) and 5% (public) by 2024, 25% and 30% by 2034, and 70% and 60% by 2100, respectively. The median member of the public thought there was a higher chance that AI systems with subjective experience would never exist (25%) than the median AI researcher did (10%). Both groups perceived a need for multidisciplinary expertise to assess AI subjective experience. Although support for welfare protections for such AI systems exceeded opposition, it remained far lower than support for protections for animals or the environment. Attitudes toward moral and governance issues were divided in both groups, especially regarding whether such systems should be created and what rights or protections they should receive. Y et a majority of respondents in both groups agreed that safeguards against the potential risks from AI systems with subjective experience should be implemented by AI developers now, and if created, AI systems with subjective experience should treat others well, behave ethically, and be held accountable. Overall, these results suggest that both AI researchers and the public regard the emergence of AI systems with subjective experience as a possibility this century, though substantial uncertainty and disagreement remain about the timeline and appropriate response. Noemi Dreksler (corresponding author) can be reached under noemi.dreksler@governance.ai.

Telegram has become a major space for political discourse and alternative media. However, its lack of moderation allows misinformation, extremism, and toxicity to spread. While prior research focused on these particular phenomena or topics, these have mostly been examined separately, and a broader understanding of the Telegram ecosystem is still missing. In this work, we fill this gap by conducting a large-scale analysis of the Italian Telegram sphere, leveraging a dataset of 186 million messages from 13,151 chats collected in 2023. Using network analysis, Large Language Models, and toxicity detection tools, we examine how different thematic communities form, align ideologically, and engage in harmful discourse within the Italian cultural context. Results show strong thematic and ideological homophily. We also identify mixed ideological communities where far-left and far-right rhetoric coexist on particular geopolitical issues. Beyond political analysis, we find that toxicity, rather than being isolated in a few extreme chats, appears widely normalized within highly toxic communities. Moreover, we find that Italian discourse primarily targets Black people, Jews, and gay individuals independently of the topic. Finally, we uncover common trend of intra-national hostility, where Italians often attack other Italians, reflecting regional and intra-regional cultural conflicts that can be traced back to old historical divisions. This study provides the first large-scale mapping of the Italian Telegram ecosystem, offering insights into ideological interactions, toxicity, and identity-targets of hate and contributing to research on online toxicity across different cultural and linguistic contexts on Telegram.

As leading examples of large language models, ChatGPT and Gemini claim to provide accurate and unbiased information, emphasizing their commitment to political neutrality and avoidance of personal bias. This research investigates the political tendency of large language models and the existence of differentiation according to the query language. For this purpose, ChatGPT and Gemini were subjected to a political axis test using 14 different languages. The findings of the study suggest that these large language models do exhibit political tendencies, with both models demonstrating liberal and leftist biases. A comparative analysis revealed that Gemini exhibited a more pronounced liberal and left-wing tendency compared to ChatGPT. The study also found that these political biases varied depending on the language used for inquiry. The study delves into the factors that constitute political tendencies and linguistic differentiation, exploring differences in the sources and scope of educational data, structural and grammatical features of languages, cultural and political contexts, and the model's response to linguistic features. From this standpoint, and an ethical perspective, it is proposed that artificial intelligence tools should refrain from asserting a lack of political tendencies and neutrality, instead striving for political neutrality and executing user queries by incorporating these tendencies.

We introduce an Item Response Theory (IRT)-based framework to detect and quantify socioeconomic bias in large language models (LLMs) without relying on subjective human judgments. Unlike traditional methods, IRT accounts for item difficulty, improving ideological bias estimation. We fine-tune two LLM families (Meta-LLaMa 3.2-1B-Instruct and Chat- GPT 3.5) to represent distinct ideological positions and introduce a two-stage approach: (1) modeling response avoidance and (2) estimating perceived bias in answered responses. Our results show that off-the-shelf LLMs often avoid ideological engagement rather than exhibit bias, challenging prior claims of partisanship. This empirically validated framework enhances AI alignment research and promotes fairer AI governance.

Large Language Models (LLMs) are increasingly used to assess news credibility, yet little is known about how they make these judgments. While prior research has examined political bias in LLM outputs or their potential for automated fact-checking, their internal evaluation processes remain largely unexamined. Understanding how LLMs assess credibility provides insights into AI behavior and how credibility is structured and applied in large-scale language models. This study benchmarks the reliability and political classifications of state-of-the-art LLMs - Gemini 1.5 Flash (Google), GPT-4o mini (OpenAI), and LLaMA 3.1 (Meta) - against structured, expert-driven rating systems such as NewsGuard and Media Bias Fact Check. Beyond assessing classification performance, we analyze the linguistic markers that shape LLM decisions, identifying which words and concepts drive their evaluations. We uncover patterns in how LLMs associate credibility with specific linguistic features by examining keyword frequency, contextual determinants, and rank distributions. Beyond static classification, we introduce a framework in which LLMs refine their credibility assessments by retrieving external information, querying other models, and adapting their responses. This allows us to investigate whether their assessments reflect structured reasoning or rely primarily on prior learned associations.

The analysis of political biases in large language models (LLMs) has primarily examined these systems as single entities with fixed viewpoints. While various methods exist for measuring such biases, the impact of persona-based prompting on LLMs' political orientation remains unexplored. In this work we leverage PersonaHub, a collection of synthetic persona descriptions, to map the political distribution of persona-based prompted LLMs using the Political Compass Test (PCT). We then examine whether these initial compass distributions can be manipulated through explicit ideological prompting towards diametrically opposed political orientations: right-authoritarian and left-libertarian. Our experiments reveal that synthetic personas predominantly cluster in the left-libertarian quadrant, with models demonstrating varying degrees of responsiveness when prompted with explicit ideological descriptors. While all models demonstrate significant shifts towards right-authoritarian positions, they exhibit more limited shifts towards left-libertarian positions, suggesting an asymmetric response to ideological manipulation that may reflect inherent biases in model training.

This work investigates the political biases and personality traits of ChatGPT, specifically comparing GPT-3.5 to GPT-4. In addition, the ability of the models to emulate political viewpoints (e.g., liberal or conservative positions) is analyzed. The Political Compass Test and the Big Five Personality Test were employed 100 times for each scenario, providing statistically significant results and an insight into the results correlations. The responses were analyzed by computing averages, standard deviations, and performing significance tests to investigate differences between GPT-3.5 and GPT-4. Correlations were found for traits that have been shown to be interdependent in human studies. Both models showed a progressive and libertarian political bias, with GPT-4's biases being slightly, but negligibly, less pronounced. Specifically, on the Political Compass, GPT-3.5 scored -6.59 on the economic axis and -6.07 on the social axis, whereas GPT-4 scored -5.40 and -4.73. In contrast to GPT-3.5, GPT-4 showed a remarkable capacity to emulate assigned political viewpoints, accurately reflecting the assigned quadrant (libertarian-left, libertarian-right, authoritarian-left, authoritarian-right) in all four tested instances. On the Big Five Personality Test, GPT-3.5 showed highly pronounced Openness and Agreeableness traits (O: 85.9%, A: 84.6%). Such pronounced traits correlate with libertarian views in human studies. While GPT-4 overall exhibited less pronounced Big Five personality traits, it did show a notably higher Neuroticism score. Assigned political orientations influenced Openness, Agreeableness, and Conscientiousness, again reflecting interdependencies observed in human studies. Finally, we observed that test sequencing affected ChatGPT's responses and the observed correlations, indicating a form of contextual memory.

Researchers and practitioners interested in computational politics rely on automatic content analysis tools to make sense of the large amount of political texts available on the Web. Such tools should provide objective and subjective aspects at different granularity levels to make the analyses useful in practice. Existing methods produce interesting insights for objective aspects, but are limited for subjective ones, are often limited to national contexts, and have limited explainability. We introduce a text analysis framework which integrates both perspectives and provides a fine-grained processing of subjective aspects. Information retrieval techniques and knowledge bases complement powerful natural language processing components to allow a flexible aggregation of results at different granularity levels. Importantly, the proposed bottom-up approach facilitates the explainability of the obtained results. We illustrate its functioning with insights on news outlets, political orientations, topics, individual entities, and demographic segments. The approach is instantiated on a large corpus of French news, but is designed to work seamlessly for other languages and countries.