GenAI agents are so hot right now - but are they secure?

In a rapidly evolving digital landscape where AI agents are becoming increasingly powerful and autonomous, security concerns have never been more pressing. As these intelligent systems gain the ability to interact with our data, make decisions on our behalf, and even create other agents, we face unprecedented security challenges that require innovative approaches.

During a recent episode of "Organic Talk on Artificial Intelligence" podcast, I had the opportunity to join Fraser Scott and Jose Lopez to explore the intersection of AI, software development, and cybersecurity. Our conversation highlighted a crucial emerging concern: as AI agents gain more capabilities and access to our systems, how do we ensure they remain secure, aligned with our intentions, and protected from exploitation?

As the founder of Growth Cyber, where we help AI startups with security and compliance, I've observed firsthand how the security challenges of agentic AI systems differ fundamentally from traditional software security. The OWASP Agentic AI Initiative, which I'm currently working on, aims to address these unique threats by developing community-driven guidance specifically for securing AI agent systems. In support of this initiative I recently helped host the OWASP Gen AI Security Project’s – NYC Insecure Agents Hackathon on April 1, 2025. One of the key takeaways from this event was how easy it was to create an insecure agent. 

Throughout our podcast discussion, we explored several critical themes that anyone working with AI agents should consider - from the challenges of "vibe coding" to the potential for agent misalignment, and from the need for runtime security monitoring to the possibility of agents policing other agents. The landscape is complex, evolving rapidly, and demands our attention.

Did you find this exploration of AI and software development helpful? Listen, like, and subscribe to our podcast Organic Talk on Artificial Intelligence for more insights on navigating the rapidly evolving landscape of artificial intelligence in technical fields. See all Podcast episodes here.