Seeing Beyond the AI Prompt: Peter Garraghan on the Urgent Future of AI Security

Send a textIn this episode of Secure Insights, we welcome Peter Garraghan, Founder, former CEO and now Chief Science Officer at Mindgard to discuss the fast-emerging risks created by today’s AI systems. AI has evolved at extraordinary speed. In just a short time, we’ve moved from basic LLMs like ChatGPT to complex, agentic AI capable of automating decisions and scaling information in ways that outpace human oversight. With this explosion of adoption comes an urgent security problem: traditional tools simply can’t detect or defend against AI-specific threats. As companies rush to integrate AI and drive profitability, they’re exposing themselves to new vulnerabilities they don’t yet understand. Peter breaks down why visibility into AI behaviour is so limited, why agentic AI is fundamentally different from earlier models, and how these gaps create an expanding attack surface that organisations can no longer ignore.