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.

Within OSNs, many of our supposedly online friends may instead be fake accounts called social bots, part of large groups that purposely re-share targeted content. Here, we study retweeting behaviors on Twitter, with the ultimate goal of detecting retweeting social bots. We collect a dataset of 10M retweets. We design a novel visualization that we leverage to highlight benign and malicious patterns of retweeting activity. In this way, we uncover a 'normal' retweeting pattern that is peculiar of human-operated accounts, and 3 suspicious patterns related to bot activities. Then, we propose a bot detection technique that stems from the previous exploration of retweeting behaviors. Our technique, called Retweet-Buster (RTbust), leverages unsupervised feature extraction and clustering. An LSTM autoencoder converts the retweet time series into compact and informative latent feature vectors, which are then clustered with a hierarchical density-based algorithm. Accounts belonging to large clusters characterized by malicious retweeting patterns are labeled as bots. RTbust obtains excellent detection results, with F1 = 0.87, whereas competitors achieve F1 < 0.76. Finally, we apply RTbust to a large dataset of retweets, uncovering 2 previously unknown active botnets with hundreds of accounts.

The cognitive research on reputation has shown several interesting properties that can improve both the quality of services and the security in distributed electronic environments. In this paper, the impact of reputation on decision-making under scarcity of information will be shown. First, a cognitive theory of reputation will be presented, then a selection of simulation experimental results from different studies will be discussed. Such results concern the benefits of reputation when agents need to find out good sellers in a virtual market-place under uncertainty and informational cheating.

The study of opinions, their formation and change, is one of the defining topics addressed by social psychology, but in recent years other disciplines, as computer science and complexity, have addressed this challenge. Despite the flourishing of different models and theories in both fields, several key questions still remain unanswered. The aim of this paper is to challenge the current theories on opinion by putting forward a cognitively grounded model where opinions are described as specific mental representations whose main properties are put forward. A comparison with reputation will be also presented.

The study of opinions, their formation and change, is one of the defining topics addressed by social psychology, but in recent years other disciplines, like computer science and complexity, have tried to deal with this issue. Despite the flourishing of different models and theories in both fields, several key questions still remain unanswered. The understanding of how opinions change and the way they are affected by social influence are challenging issues requiring a thorough analysis of opinion per se but also of the way in which they travel between agents' minds and are modulated by these exchanges. To account for the two-faceted nature of opinions, which are mental entities undergoing complex social processes, we outline a preliminary model in which a cognitive theory of opinions is put forward and it is paired with a formal description of them and of their spreading among minds. Furthermore, investigating social influence also implies the necessity to account for the way in which people change their minds, as a consequence of interacting with other people, and the need to explain the higher or lower persistence of such changes.

Despite the increasing diffusion of the Internet technology, TV remains the principal medium of communication. People's perceptions, knowledge, beliefs and opinions about matter of facts get (in)formed through the information reported on by the mass-media. However, a single source of information (and consensus) could be a potential cause of anomalies in the structure and evolution of a society. Hence, as the information available (and the way it is reported) is fundamental for our perceptions and opinions, the definition of conditions allowing for a good information to be disseminated is a pressing challenge. In this paper starting from a report on the last Italian political campaign in 2008, we derive a socio-cognitive computational model of opinion dynamics where agents get informed by different sources of information. Then, a what-if analysis, performed trough simulations on the model's parameters space, is shown. In particular, the scenario implemented includes three main streams of information acquisition, differing in both the contents and the perceived reliability of the messages spread. Agents' internal opinion is updated either by accessing one of the information sources, namely media and experts, or by exchanging information with one another. They are also endowed with cognitive mechanisms to accept, reject or partially consider the acquired information.