Generating Threat Modeling Antipersonas with Large Language Models: A Feasibility Study
Resumo
Early identification of security threats during software requirements engineering is a critical challenge, often hindered by limited security expertise and time constraints. This work investigates the ability of LLMs to automatically generate antipersonas from software requirements. We propose a structured antipersona representation and a prompt-based approach grounded in STRIDE and MITRE ATT&CK frameworks, to guide LLMs in generating antipersonas, and we conducted a preliminary feasibility study to evaluate the structural completeness, specificity, and adherence to these frameworks of the generated antipersonas. The generated antipersonas were predominantly assessed as structurally complete, specific, and largely adherent to STRIDE and MITRE ATT&CK, though adherence to STRIDE showed low inter-rater agreement, and we also identified recurring limitations in framework adherence. The findings suggest that LLMs are feasible for generating well-structured antipersonas that can support, as a preliminary step, early threat identification during requirements engineering, though careful human validation against real-world threats remains necessary to ensure the practical relevance of the generated antipersonas.
Palavras-chave:
Threat Modeling, Antipersona, Large Language Models, Prompt Engineering, Security Requirements
Referências
Anthropic. 2026. Introducing Claude Sonnet 4.6. Anthropic News. [link]
Gheventer at al. 2026. Investigating the Use of Large Language Models for Generating Abuser Stories for Early Security Threat Identification.
Andrea S. Atzeni, Cesare Cameroni, Shamal Faily, John Lyle, and Ivan Fléchais. 2011. Here’s Johnny: A Methodology for Developing Attacker Personas. In 2011 Sixth International Conference on Availability, Reliability and Security (ARES). IEEE, 722–727. DOI: 10.1109/ARES.2011.115
Martin Bedoya, Sara Palacios, Daniel Díaz-López, Estefania Laverde, and Pantaleone Nespoli. 2024. Enhancing DevSecOps practice with Large Language Models and Security Chaos Engineering. International Journal of Information Security 23 (2024), 3765–3788. DOI: 10.1007/s10207-024-00909-w
Tom B. Brown, Benjamin Mann, Nick Ryder, et al. 2020. Language Models are Few-Shot Learners. In Advances in Neural Information Processing Systems 33 (NeurIPS 2020).
Badhan Chandra Das, M. Hadi Amini, and Yanzhao Wu. 2025. Security and Privacy Challenges of Large Language Models: A Survey. Comput. Surveys 57, 6 (2025), 1–39. DOI: 10.1145/3712001
Isra Elsharef, Zhen Zeng, and Zhongshu Gu. 2024. Facilitating Threat Modeling by Leveraging Large Language Models. Proceedings 2024 Workshop on AI Systems with Confidential Computing (AISCC) (2024). DOI: 10.14722/aiscc.2024.23016
Mavroeidis et al. 2021. Threat Actor Type Inference and Characterization within Cyber Threat Intelligence. In 2021 13th International Conference on Cyber Conflict (CyCon). IEEE, 327–352. DOI: 10.23919/CYCON51939.2021.9468305
Schuller et al. 2024. Generating Personas Using LLMs and Assessing Their Viability. In Extended Abstracts of the 2024 CHI Conf. on Human Factors in Computing Systems (CHI EA ’24). ACM, Article 179, 7 pages. DOI: 10.1145/3613905.3650860
Sahoo et al. 2024. A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications. arXiv:2402.07927
Wimbauer et al. 2025. ThreatCompute: Leveraging LLMs for Automated Threat Modeling of Cloud-Native Applications. In Proceedings of the 2025 Cloud Computing Security Workshop (CCSW ’25). ACM, 14–27. DOI: 10.1145/3733812.3765533
Shawn Hernan, Scott Lambert, Tomasz Ostwald, and Adam Shostack. 2006. Uncover Security Design Flaws Using The STRIDE Approach. MSDN Magazine (November 2006).
J Richard Landis and Gary G Koch. 1977. The measurement of observer agreement for categorical data. biometrics (1977), 159–174.
Adam Steele and Xiaoli Jia. 2008. Adversary Centered Design: Threat Modeling Using Anti-Scenarios, Anti-Use Cases and Anti-Personas. In Proc. of the International Conference on Information and Knowledge Engineering (IKE). 367–370.
Blake E. Strom, Andy Applebaum, Doug P. Miller, Kathryn C. Nickels, Adam G. Pennington, and Cody B. Thomas. 2018. MITRE ATT&CK: Design and Philosophy. Technical Report. The MITRE Corporation.
Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, and Denny Zhou. 2022. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. In Advances in Neural Information Processing Systems 35 (NeurIPS 2022). DOI: 10.5555/3600270.3602070
Gheventer at al. 2026. Investigating the Use of Large Language Models for Generating Abuser Stories for Early Security Threat Identification.
Andrea S. Atzeni, Cesare Cameroni, Shamal Faily, John Lyle, and Ivan Fléchais. 2011. Here’s Johnny: A Methodology for Developing Attacker Personas. In 2011 Sixth International Conference on Availability, Reliability and Security (ARES). IEEE, 722–727. DOI: 10.1109/ARES.2011.115
Martin Bedoya, Sara Palacios, Daniel Díaz-López, Estefania Laverde, and Pantaleone Nespoli. 2024. Enhancing DevSecOps practice with Large Language Models and Security Chaos Engineering. International Journal of Information Security 23 (2024), 3765–3788. DOI: 10.1007/s10207-024-00909-w
Tom B. Brown, Benjamin Mann, Nick Ryder, et al. 2020. Language Models are Few-Shot Learners. In Advances in Neural Information Processing Systems 33 (NeurIPS 2020).
Badhan Chandra Das, M. Hadi Amini, and Yanzhao Wu. 2025. Security and Privacy Challenges of Large Language Models: A Survey. Comput. Surveys 57, 6 (2025), 1–39. DOI: 10.1145/3712001
Isra Elsharef, Zhen Zeng, and Zhongshu Gu. 2024. Facilitating Threat Modeling by Leveraging Large Language Models. Proceedings 2024 Workshop on AI Systems with Confidential Computing (AISCC) (2024). DOI: 10.14722/aiscc.2024.23016
Mavroeidis et al. 2021. Threat Actor Type Inference and Characterization within Cyber Threat Intelligence. In 2021 13th International Conference on Cyber Conflict (CyCon). IEEE, 327–352. DOI: 10.23919/CYCON51939.2021.9468305
Schuller et al. 2024. Generating Personas Using LLMs and Assessing Their Viability. In Extended Abstracts of the 2024 CHI Conf. on Human Factors in Computing Systems (CHI EA ’24). ACM, Article 179, 7 pages. DOI: 10.1145/3613905.3650860
Sahoo et al. 2024. A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications. arXiv:2402.07927
Wimbauer et al. 2025. ThreatCompute: Leveraging LLMs for Automated Threat Modeling of Cloud-Native Applications. In Proceedings of the 2025 Cloud Computing Security Workshop (CCSW ’25). ACM, 14–27. DOI: 10.1145/3733812.3765533
Shawn Hernan, Scott Lambert, Tomasz Ostwald, and Adam Shostack. 2006. Uncover Security Design Flaws Using The STRIDE Approach. MSDN Magazine (November 2006).
J Richard Landis and Gary G Koch. 1977. The measurement of observer agreement for categorical data. biometrics (1977), 159–174.
Adam Steele and Xiaoli Jia. 2008. Adversary Centered Design: Threat Modeling Using Anti-Scenarios, Anti-Use Cases and Anti-Personas. In Proc. of the International Conference on Information and Knowledge Engineering (IKE). 367–370.
Blake E. Strom, Andy Applebaum, Doug P. Miller, Kathryn C. Nickels, Adam G. Pennington, and Cody B. Thomas. 2018. MITRE ATT&CK: Design and Philosophy. Technical Report. The MITRE Corporation.
Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Brian Ichter, Fei Xia, Ed Chi, Quoc Le, and Denny Zhou. 2022. Chain-of-Thought Prompting Elicits Reasoning in Large Language Models. In Advances in Neural Information Processing Systems 35 (NeurIPS 2022). DOI: 10.5555/3600270.3602070
Publicado
11/09/2026
Como Citar
ALVES, Anna Cristina Ferreira; MELLO, Rafael Maiani de.
Generating Threat Modeling Antipersonas with Large Language Models: A Feasibility Study. In: WORKSHOP BRASILEIRO DE ENGENHARIA DE SOFTWARE INTELIGENTE (ISE), 5. , 2026, São Paulo/SP.
Anais [...].
Porto Alegre: Sociedade Brasileira de Computação,
2026
.
p. 95-100.
DOI: https://doi.org/10.5753/ise.2026.33028.
