Software security in the era of autonomous and generative AI

The field of software security has become increasingly intertwined with advances in artificial intelligence (AI), especially as generative and autonomous AI systems are integrated into software infrastructure at unprecedented rates. While such integration brings revolutionary capabilities—featuring intelligent automation, adaptive threat detection, and dynamic system responses—it also creates novel security challenges. Contemporary research highlights how adversaries exploit vulnerabilities in software supporting large language models (LLMs), agent-based architectures, and automated decision-making, launching attacks such as data poisoning, prompt injection, and model theft. This evolving landscape exposes gaps in conventional software security measures, raising questions about code trustworthiness, model integrity, and the limits of current development practices. Notably, studies have underlined the dual role of AI: as both a tool for improving security and a source of new risks, necessitating fresh approaches to the core principles of secure software design and deployment. Despite ongoing advancements, much of existing research remains fragmented, with a lack of holistic frameworks to address the compounded threats introduced by AI-powered software. Recent findings indicate the need for new architectural strategies, secure machine learning pipelines, and advanced adversarial testing methods. Moreover, rising concerns over explainability, human oversight, and regulatory compliance have drawn attention to the importance of designing trustworthy software capable of robust operation in adversarial and unpredictable environments. However, significant gaps persist in cross-disciplinary solutions that integrate governance, human factors, and technical safeguards within the software lifecycle. This Research Topic aims to unite the fields of software security and AI by exploring both the challenges and solutions at their intersection. The primary goal is to stimulate the development of secure, resilient, and trustworthy software systems in the age of autonomous and generative AI. Key objectives include analyzing the evolving threat landscape, establishing robust defense mechanisms, and designing explainable, human-centric software architectures. The initiative seeks to address open questions around secure software development practices for AI-enhanced systems, methods for mitigating adversarial threats targeting software and models, and frameworks for accountable deployment. By fostering collaboration across disciplines—spanning machine learning, systems and software engineering, ethics, and governance—this Research Topic aspires to set the groundwork for new paradigms in software security. The scope of this Research Topic encompasses research at the intersection of AI and software security, with a focus on practical, theoretical, and interdisciplinary contributions. We encourage submissions that generate actionable insights, propose innovative frameworks, and rigorously assess the security of AI-driven software. To gather further insights on secure software in the AI era, we welcome articles addressing, but not limited to, the following themes: - Secure software engineering for systems integrating AI components - Vulnerability discovery and mitigation in AI-powered software - Resilience and robustness of software against AI-driven cyber threats - Secure development and deployment of autonomous and generative AI models - Software supply chain security in machine learning environments - Explainable, auditable, and trustworthy software design - Human factors, usability, and organizational practices in AI software security - Regulatory, ethical, and governance aspects of secure AI-enabled software - Integrating secure software practices with emerging paradigms like Zero Trust and post-quantum computing The Research Topic welcomes original research, review articles, perspectives, and case studies. Interdisciplinary and practice-bridging works are particularly encouraged. Topic editors: Tanja Pavleska, Haipeng Cai Source: https://www.frontiersin.org/research-topics/79870/software-security-in-the-era-of-autonomous-and-generative-ai