{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2025,12,11]],"date-time":"2025-12-11T06:35:20Z","timestamp":1765434920547,"version":"3.46.0"},"reference-count":68,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,12,11]],"date-time":"2025-12-11T00:00:00Z","timestamp":1765411200000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Comput. Sci."],"abstract":"<jats:sec>\n                    <jats:title>Introduction<\/jats:title>\n                    <jats:p>Game theory has long served as a foundational tool in cybersecurity to test, predict, and design strategic interactions between attackers and defenders. The recent advent of Large Language Models (LLMs) offers new tools and challenges for the security of computer systems. In this work, we investigate whether classical game-theoretic frameworks can effectively capture the behaviors of LLM-driven actors and bots.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Methods<\/jats:title>\n                    <jats:p>Using a reproducible framework for game-theoretic LLM agents, we investigate two canonical scenarios\u2014the one-shot zero-sum game and the dynamic Prisoner's Dilemma\u2014and we test whether LLMs converge to expected outcomes or exhibit deviations due to embedded biases. We experiments on four state-of-the-art LLMs and five natural languages (English, French, Arabic, Vietnamese, and Mandarin Chinese) to assess linguistic sensitivity.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Results<\/jats:title>\n                    <jats:p>For both games, we observe that the final payoffs are influenced by agents characteristics such as personality traits or knowledge of repeated rounds. We also uncover an unexpected sensitivity of the final payoffs to the choice of languages, which should warn against indiscriminate application of LLMs in cybersecurity applications and call for in-depth studies, as LLMs may behave differently when deployed in different countries. We also employ quantitative metrics to evaluate the internal consistency and cross-language stability of LLM agents.<\/jats:p>\n                  <\/jats:sec>\n                  <jats:sec>\n                    <jats:title>Discussion<\/jats:title>\n                    <jats:p>In addition to uncovering unexpected behaviors requiring attention by scholars and practitioners, our work can help guide the selection of the most stable LLMs and optimizing models for secure applications.<\/jats:p>\n                  <\/jats:sec>","DOI":"10.3389\/fcomp.2025.1703586","type":"journal-article","created":{"date-parts":[[2025,12,11]],"date-time":"2025-12-11T06:32:24Z","timestamp":1765434744000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":0,"title":["Can LLMs effectively provide game-theoretic-based scenarios for cybersecurity?"],"prefix":"10.3389","volume":"7","author":[{"given":"Daniele","family":"Proverbio","sequence":"first","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Alessio","family":"Buscemi","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Alessandro","family":"Di Stefano","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"The Anh","family":"Han","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"German","family":"Castignani","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]},{"given":"Pietro","family":"Li\u00f2","sequence":"additional","affiliation":[],"role":[{"role":"author","vocabulary":"crossref"}]}],"member":"1965","published-online":{"date-parts":[[2025,12,11]]},"reference":[{"key":"B1","doi-asserted-by":"publisher","first-page":"26","DOI":"10.1109\/MSP.2005.22","article-title":"Privacy and rationality in individual decision making","volume":"3","author":"Acquisti","year":"2005","journal-title":"IEEE Secur. Priv"},{"key":"B2","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1038\/s41562-025-02172-y","article-title":"Playing repeated games with large language models","volume":"2025","author":"Akata","year":"2025","journal-title":"Nat. Hum. Behav"},{"key":"B3","doi-asserted-by":"publisher","first-page":"129627","DOI":"10.1016\/j.amc.2025.129627","article-title":"Trust ai regulation? discerning users are vital to build trust and effective ai regulation","volume":"508","author":"Alalawi","year":"2026","journal-title":"Appl. Math. Comput"},{"key":"B4","doi-asserted-by":"publisher","first-page":"881","DOI":"10.1007\/s13235-019-00335-x","article-title":"Preface to the focused issue on dynamic games in cyber security","volume":"9","author":"Amin","year":"2019","journal-title":"Dyn. Games Applic"},{"key":"B5","first-page":"60","article-title":"\u201cA zero-sum power allocation game in the parallel gaussian wiretap channel with an unfriendly jammer,\u201d","volume-title":"2012 IEEE ICCS","author":"Ara","year":"2012"},{"key":"B6","doi-asserted-by":"crossref","first-page":"1009","DOI":"10.1109\/ICSADL65848.2025.10933026","article-title":"\u201cEvolving strategies: LLMS as game players,\u201d","volume-title":"2025 4th International Conference on Sentiment Analysis and Deep Learning (ICSADL)","author":"Avinash","year":"2025"},{"key":"B7","article-title":"Media and responsible ai governance: a game-theoretic and LLM analysis","author":"Balabanova","year":"2025","journal-title":"arXiv:2503.09858"},{"key":"B8","article-title":"Co-evolutionary dynamics of attack and defence in cybersecurity","author":"Bashir","year":"2025","journal-title":"arXiv preprint arXiv:2505.19338"},{"key":"B9","doi-asserted-by":"publisher","first-page":"25","DOI":"10.5711\/1082598316425","article-title":"A game-theoretic model for defense of an oceanic bastion against submarines","volume":"16","author":"Brown","year":"2011","journal-title":"Milit. Oper. Res"},{"key":"B10","article-title":"Large language models' detection of political orientation in newspapers","author":"Buscemi","year":"2024","journal-title":"arXiv preprint arXiv:2406.00018"},{"key":"B11","article-title":"Do LLMs trust AI regulation? Emerging behaviour of game-theoretic LLM agents","author":"Buscemi","year":"","journal-title":"arXiv:2504.08640"},{"key":"B12","article-title":"Fairgame: a framework for ai agents bias recognition using game theory","author":"Buscemi","year":"","journal-title":"arXiv preprint arXiv:2504.14325"},{"key":"B13","article-title":"\u201cCan media act as a soft regulator of safe ai development? a game theoretical analysis,\u201d","author":"Correia da Fonseca","year":"2025","journal-title":"ALIFE 2025"},{"key":"B14","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1145\/3057268","article-title":"Game theory for cyber security and privacy","volume":"50","author":"Do","year":"2017","journal-title":"ACM Comput. Surv"},{"key":"B15","doi-asserted-by":"publisher","DOI":"10.52202\/079017-0885","article-title":"\u201cGtbench: uncovering the strategic reasoning capabilities of llms via game-theoretic evaluations,\u201d","author":"Duan","year":"2024","journal-title":"Advances in Neural Information Processing Systems"},{"key":"B16","doi-asserted-by":"publisher","first-page":"884","DOI":"10.1007\/s13235-018-00291-y","article-title":"Dynamic games in cyber-physical security: an overview","volume":"9","author":"Etesami","year":"2019","journal-title":"Dyn. Games Applic"},{"key":"B17","doi-asserted-by":"publisher","DOI":"10.1609\/aaai.v38i16.29751","article-title":"\u201cCan large language models serve as rational players in game theory? A systematic analysis,\u201d","author":"Fan","year":"2024","journal-title":"Proceedings of the AAAI Conference on Artificial Intelligence"},{"key":"B18","doi-asserted-by":"publisher","DOI":"10.2139\/ssrn.4853709","article-title":"Generative ai and large language models for cyber security: All insights you need","author":"Ferrag","year":"2024","journal-title":"Available at SSRN 4853709"},{"key":"B19","article-title":"Nicer than humans: how do large language models behave in the prisoner's dilemma?","author":"Fontana","year":"2024","journal-title":"arXiv:2406.13605"},{"key":"B20","doi-asserted-by":"publisher","first-page":"102","DOI":"10.3389\/frobt.2018.00102","article-title":"No strategy can win in the repeated prisoner's dilemma: linking game theory and computer simulations","volume":"5","author":"Garc\u00eda","year":"2018","journal-title":"Front. Robot. AI"},{"key":"B21","article-title":"\u201cConsiderations for evaluating large language models for cybersecurity tasks,\u201d","author":"Gennari","year":"2024","journal-title":"SEI Insights"},{"key":"B22","article-title":"Multi-agent risks from advanced AI","author":"Hammond","year":"2025","journal-title":"arXiv preprint arXiv:2502.14143"},{"key":"B23","article-title":"LLM multi-agent systems: challenges and open problems","author":"Han","year":"2024","journal-title":"arXiv preprint arXiv:2402.03578"},{"key":"B24","doi-asserted-by":"publisher","first-page":"881","DOI":"10.1613\/jair.1.12225","article-title":"To regulate or not: a social dynamics analysis of an idealised ai race","volume":"69","author":"Han","year":"2020","journal-title":"J. Artif. Intell. Res"},{"key":"B25","doi-asserted-by":"publisher","first-page":"17","DOI":"10.1111\/0272-4332.t01-1-00002","article-title":"Probabilistic risk analysis and game theory","volume":"22","author":"Hausken","year":"2002","journal-title":"Risk Anal"},{"key":"B26","doi-asserted-by":"publisher","first-page":"100204","DOI":"10.1016\/j.iot.2020.100204","article-title":"Cyber resilience in firms, organizations and societies","volume":"11","author":"Hausken","year":"2020","journal-title":"Internet Things"},{"key":"B27","doi-asserted-by":"publisher","first-page":"355","DOI":"10.1016\/j.ejor.2023.12.023","article-title":"Fifty years of operations research in defense","volume":"318","author":"Hausken","year":"2024","journal-title":"Eur. J. Oper. Res"},{"key":"B28","doi-asserted-by":"publisher","first-page":"28","DOI":"10.3390\/g15040028","article-title":"A review of attacker-defender games and cyber security","volume":"15","author":"Hausken","year":"2024","journal-title":"Games"},{"key":"B29","doi-asserted-by":"publisher","first-page":"122","DOI":"10.1109\/MWC.007.2400133","article-title":"Generative ai for game theory-based mobile networking","volume":"32","author":"He","year":"2025","journal-title":"IEEE Wirel. Commun"},{"key":"B30","article-title":"\u201cLarge language models are bad game theoretic reasoners: Evaluating performance and bias in two-player non-zero-sum games,\u201d","author":"Herr","year":"2024","journal-title":"ICML 2024 Workshop on LLMs and Cognition"},{"key":"B31","article-title":"\u201cCompeting large language models in multi-agent gaming environments,\u201d","author":"Huang","year":"2025","journal-title":"13th International Conference on Learning Representations"},{"key":"B32","doi-asserted-by":"publisher","first-page":"1242","DOI":"10.1109\/TMC.2010.87","article-title":"A belief evaluation framework in autonomous manets under noisy and imperfect observation: Vulnerability analysis and cooperation enforcement","volume":"9","author":"Ji","year":"2010","journal-title":"IEEE Trans. Mobile Comput"},{"key":"B33","article-title":"Large language model strategic reasoning evaluation through behavioral game theory","author":"Jia","year":"2025","journal-title":"arXiv preprint arXiv:2502.20432"},{"key":"B34","doi-asserted-by":"crossref","first-page":"1973","DOI":"10.1109\/GLOCOMW.2010.5700289","article-title":"\u201cGame theoretic analysis of cooperation in autonomous multi hop networks: The consequences of unequal traffic load,\u201d","volume-title":"2010 IEEE Globecom Workshops","author":"Kamhoua","year":"2010"},{"key":"B35","first-page":"360","article-title":"\u201cTesting for hardware trojans: a game-theoretic approach,\u201d","volume-title":"International Conference on Decision and Game Theory for Security","author":"Kamhoua","year":"2014"},{"key":"B36","doi-asserted-by":"publisher","first-page":"30","DOI":"10.3390\/computation13020030","article-title":"From vulnerability to defense: the role of large language models in enhancing cybersecurity","volume":"13","author":"Kasri","year":"2025","journal-title":"Computation"},{"key":"B37","doi-asserted-by":"publisher","DOI":"10.1109\/WCS.2013.7050508","article-title":"\u201cThe digital prisoner's dilemma: challenges and opportunities for cooperation,\u201d","author":"Kostyuk","year":"2013","journal-title":"2013 World Cyberspace Cooperation Summit IV (WCC4)"},{"key":"B38","doi-asserted-by":"publisher","first-page":"100779","DOI":"10.1016\/j.patter.2023.100779","article-title":"GPT detectors are biased against non-native english writers","volume":"4","author":"Liang","year":"2023","journal-title":"Patterns"},{"key":"B39","doi-asserted-by":"crossref","first-page":"44","DOI":"10.1007\/978-3-642-38033-4_4","article-title":"\u201cOptimal defense strategies for ddos defender using bayesian game model,\u201d","volume-title":"9th Conference on Information Security Practice and Experience (ISPEC)","author":"Liu","year":"2013"},{"key":"B40","doi-asserted-by":"publisher","first-page":"18490","DOI":"10.1038\/s41598-024-69032-z","article-title":"Strategic behavior of large language models and the role of game structure vs. contextual framing","volume":"14","author":"Lor\u00e9","year":"2024","journal-title":"Sci. Rep"},{"key":"B41","doi-asserted-by":"publisher","first-page":"283","DOI":"10.1016\/j.plrev.2024.10.013","article-title":"LLMS and generative agent-based models for complex systems research","volume":"51","author":"Lu","year":"2024","journal-title":"Phys. Life Rev"},{"key":"B42","doi-asserted-by":"publisher","first-page":"9","DOI":"10.1023\/A:1009905800005","article-title":"Quantal response equilibria for extensive form games","volume":"1","author":"McKelvey","year":"1998","journal-title":"Exper. Econ"},{"key":"B43","article-title":"Cybercrime to cost the world $10.5 trillion annually by 2025.","author":"Morgan","year":"2020","journal-title":"Cybercrime Magazine"},{"key":"B44","doi-asserted-by":"crossref","first-page":"6126","DOI":"10.1109\/CDC51059.2022.9992468","article-title":"\u201cA zero-sum game framework for optimal sensor placement in uncertain networked control systems under cyber-attacks,\u201d","volume-title":"2022 IEEE 61st Conference on Decision and Control (CDC)","author":"Nguyen","year":"2022"},{"key":"B45","volume-title":"Game Theory","author":"Owen","year":"2013"},{"key":"B46","doi-asserted-by":"publisher","first-page":"172","DOI":"10.1287\/deca.2018.0387","article-title":"Information sharing in cybersecurity: a review","volume":"16","author":"Pala","year":"2019","journal-title":"Dec. Anal"},{"key":"B47","first-page":"1","article-title":"Leveraging predictive modeling, machine learning personalization, nlp customer support, and ai chatbots to increase customer loyalty","volume":"3","author":"Patel","year":"2020","journal-title":"Empir. Quests Manage. Essenc"},{"key":"B48","unstructured":"Petrosyan\n              A.\n            \n          \n          Estimated cost of cybercrime worldwide 2018\u20132029\n          \n          2024"},{"key":"B49","doi-asserted-by":"crossref","first-page":"1","DOI":"10.1109\/CyberSA49311.2020.9139644","article-title":"\u201cBeyond the prisoner's dilemma: the social dilemmas of cybersecurity,\u201d","volume-title":"2020 International Conference on Cyber Situational Awareness, Data Analytics and Assessment (CyberSA)","author":"Schoenherr","year":"2020"},{"key":"B50","doi-asserted-by":"publisher","DOI":"10.1145\/1852666.1852704","article-title":"\u201cGame theory for cyber security,\u201d","author":"Shiva","year":"2010","journal-title":"Proceedings of the Sixth Annual Workshop on Cyber Security and Information Intelligence Research"},{"key":"B51","doi-asserted-by":"publisher","first-page":"856","DOI":"10.1109\/TCST.2022.3207671","article-title":"A robust stackelberg game for cyber-security investment in networked control systems","volume":"31","author":"Shukla","year":"2022","journal-title":"IEEE Trans. Control Syst. Technol"},{"key":"B52","doi-asserted-by":"publisher","first-page":"39","DOI":"10.1016\/j.cose.2013.03.014","article-title":"A game theoretic defence framework against dos\/ddos cyber attacks","volume":"38","author":"Spyridopoulos","year":"2013","journal-title":"Comput. Secur"},{"key":"B53","first-page":"808","article-title":"\u201cCooperation in wireless ad hoc networks,\u201d","volume-title":"IEEE INFOCOM 2003","author":"Srinivasan","year":"2003"},{"key":"B54","article-title":"Game theory meets large language models: a systematic survey","author":"Sun","year":"2025","journal-title":"arXiv preprint arXiv:2502.09053"},{"key":"B55","article-title":"The impact of revealing large language model stochasticity on trust, reliability, and anthropomorphization","author":"Swoopes","year":"2025","journal-title":"arXiv preprint arXiv:2503.16114"},{"key":"B56","doi-asserted-by":"crossref","first-page":"1822","DOI":"10.1109\/Allerton.2012.6483443","article-title":"\u201cGame theory for security: key algorithmic principles, deployed systems, lessons learned,\u201d","volume-title":"2012 50th Annual Allerton Conference on Communication, Control, and Computing (Allerton)","author":"Tambe","year":"2012"},{"key":"B57","doi-asserted-by":"publisher","first-page":"eadq2852","DOI":"10.1126\/science.adq2852","article-title":"AI can help humans find common ground in democratic deliberation","volume":"386","author":"Tessler","year":"2024","journal-title":"Science"},{"key":"B58","doi-asserted-by":"publisher","DOI":"10.1515\/9781400829460","article-title":"\u201cTheory of games and economic behavior: 60th anniversary commemorative edition,\u201d","author":"Von Neumann","year":"2007","journal-title":"Theory of Games and Economic Behavior"},{"key":"B59","first-page":"631","article-title":"\u201cA survey of game theoretic methods for cyber security,\u201d","volume-title":"2016 IEEE 1st International Conferences on Data Science in Cyberspace (DSC)","author":"Wang","year":"2016"},{"key":"B60","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1016\/j.plrev.2015.04.033","article-title":"Universal scaling for the dilemma strength in evolutionary games","volume":"14","author":"Wang","year":"2015","journal-title":"Phys. Life Rev"},{"key":"B61","article-title":"Large language models overcome the machine penalty when acting fairly but not when acting selfishly or altruistically","author":"Wang","year":"2024","journal-title":"arXiv:2410.03724"},{"key":"B62","doi-asserted-by":"publisher","first-page":"1017","DOI":"10.1613\/jair.1.16146","article-title":"Empirical game theoretic analysis: a survey","volume":"82","author":"Wellman","year":"2025","journal-title":"J. Artif. Intell. Res"},{"key":"B63","article-title":"Towards agentic ai networking in 6G: a generative foundation model-as-agent approach","author":"Xiao","year":"2025","journal-title":"arXiv preprint arXiv:2503.15764"},{"key":"B64","doi-asserted-by":"publisher","first-page":"143806","DOI":"10.1109\/ACCESS.2024.3468914","article-title":"Applications of LLMS for generating cyber security exercise scenarios","volume":"12","author":"Yamin","year":"2024","journal-title":"IEEE Access"},{"key":"B65","doi-asserted-by":"publisher","first-page":"191","DOI":"10.1109\/MNET.2024.3521887","article-title":"Exploring equilibrium strategies in network games with generative AI","volume":"39","author":"Yang","year":"2024","journal-title":"IEEE Netw"},{"key":"B66","article-title":"Toward a multi-echelon cyber warfare theory: a meta-game-theoretic paradigm for defense and dominance","author":"Yang","year":"2025","journal-title":"arXiv preprint arXiv:2509.08976"},{"key":"B67","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1186\/s42400-025-00361-w","article-title":"When llms meet cybersecurity: a systematic literature review","volume":"8","author":"Zhang","year":"2025","journal-title":"Cybersecurity"},{"key":"B68","doi-asserted-by":"crossref","first-page":"4066","DOI":"10.1109\/CDC.2011.6161031","article-title":"\u201cRobust and resilient control design for cyber-physical systems with an application to power systems,\u201d","volume-title":"2011 50th IEEE Conference on Decision and Control and European Control Conference","author":"Zhu","year":"2011"}],"container-title":["Frontiers in Computer Science"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2025.1703586\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,12,11]],"date-time":"2025-12-11T06:32:31Z","timestamp":1765434751000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fcomp.2025.1703586\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,12,11]]},"references-count":68,"alternative-id":["10.3389\/fcomp.2025.1703586"],"URL":"https:\/\/doi.org\/10.3389\/fcomp.2025.1703586","relation":{},"ISSN":["2624-9898"],"issn-type":[{"value":"2624-9898","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,12,11]]},"article-number":"1703586"}}