The Hidden Security Risks of AI And How to Prevent Them with Donato Capitella

AI agents are rapidly entering the enterprise — but with new power comes new risk. In this episode, Kenny sits down with Donato Capitella, a principal security consultant and ethical hacker with 14+ years of experience breaking into systems before the bad actors do. Together, they unpack the real security challenges behind AI agents and GenAI adoption.Donato explains why prompt injection has become one of the most common attack vectors, how AI-powered workflows can unintentionally leak sensitive data, and why organisations should treat every LLM output as untrusted. Drawing on his experience reviewing 50+ enterprise GenAI implementations, he shares what teams consistently overlook and how leaders can design safer, more resilient AI systems before they hit production.It’s a practical, grounded conversation for COOs, CTOs, and product leaders looking to strengthen security and deploy AI agents safely without slowing innovation.Mentioned in the podcast:The LLM Application Security Canvas: https://reversec.com/service/generative-ai-security/Chapters:00:00 – Introduction to AI and Security05:04 – Understanding Ethical Hacking and Generative AI08:04 – Security Challenges in AI Implementation10:08 – The Concept of Prompt Injection and Its Risks16:03 – Guardrails for AI Implementation24:00 – Biggest Opportunities in AI31:33 – Advice for Leaders on AI Use CasesFind us on socials:🔗 Follow Kenny on LinkedIn: https://www.linkedin.com/in/kennyalegbe/🤖 Follow Brim on LinkedIn: https://www.linkedin.com/company/joinbrim/📩 For business or guest inquiries: [email protected]