Overview
Generative Artificial Intelligence (GenAI) is transforming the landscape of digital content creation—from software and code generation to text, images, and synthetic data. These technologies enable automation at scale and open new opportunities across sectors such as software engineering, cybersecurity, healthcare, and education. However, as GenAI systems become increasingly deployed in mission-critical and sensitive domains, their inherent vulnerabilities raise pressing concerns regarding security, dependability, and ethical use.
DSGenAI-2025 is an international workshop dedicated to exploring the challenges and advancements in building dependable and secure GenAI systems. The workshop will bring together researchers, practitioners, and policymakers from diverse disciplines to examine the threats and risks posed by GenAI technologies and develop strategies to improve their robustness, reliability, and trustworthiness.


We invite original research papers, position papers, tool demonstrations, and case studies on topics including, but not limited to:
• Secure training and fine-tuning of generative AI models to prevent adversarial manipulation and backdoor attacks.
• Adversarial attacks and defenses against GenAI models and outputs, including evasion, poisoning, and prompt injection techniques.
• Dependability and fault tolerance in GenAI pipelines, focusing on robust model performance in dynamic or degraded environments.
• Explainability and interpretability of AI-generated content to support human oversight and trust.
• Secure prompt engineering, mitigation of prompt injection, prompt leakage, and malicious output risks.
• Formal methods for verification and validation of AI-generated artifacts, especially code and scripts.
• Privacy-preserving GenAI techniques, including federated learning, data minimization, and synthetic data generation.
• Ethical, legal, and regulatory compliance in GenAI system development and deployment.
• Benchmarking and evaluation metrics for assessing GenAI system security, safety, and dependability.