How to Attack AI Systems and How to Defend Them

Learn how attackers compromise AI systems and discover effective defensive strategies in this 42-minute conference talk from DevConf.IN 2026. Explore the expanding attack surface of modern machine learning pipelines, from data collection and training to inference, through real research examples and case studies. Examine various adversarial manipulation techniques that threaten AI systems and understand how these attacks work in practice. Discover defensive strategies grounded in current academic and industry research, focusing on practical approaches that organizations can implement today. Gain insights into architecting AI systems that maintain trustworthiness even under adversarial pressure, with emphasis on realistic and effective security measures rather than theoretical concepts.