As generative AI becomes embedded in everyday software development, organizations are moving faster than ever, but often without full visibility into the security consequences. As generative AI becomes a mainstream tool for software development, serious concerns around vulnerabilities, governance, and accountability are arising. With development velocity accelerating and attack surfaces expanding, leaders are now facing a critical question: can we trust AI to write secure code?

Recent large-scale evaluations of AI models across common programming languages reveal unsettling patterns, including frequent failures on secure coding tasks, recurring vulnerability types, and a lack of meaningful security improvement as models evolve. These findings challenge assumptions that newer or larger models are inherently safer, and they highlight blind spots that many development and security teams may not yet be equipped to manage. Understanding where AI succeeds, where it falls short, and how risk quietly enters the software lifecycle is becoming essential.

Join Veracode and a select group of your peers as we explore what these trends mean for developers, security teams, and the broader business. Together, we’ll examine the real-world implications of AI-driven development, the hidden risks shaping today’s software supply chains, and how organizations are rethinking trust, accountability, and security in an AI-accelerated world.