Can My AI Be Hacked?

In this episode of Unspoken Security, host AJ Nash speaks with Dr. Peter Garraghan, CEO and CTO of Mindgard. They discuss the real-world security risks of artificial intelligence. Peter starts with a simple point: AI is just software, and software is easy to break. He urges businesses using AI to step back and truly understand its vulnerabilities.Peter draws parallels between the current AI boom and past technology cycles like cloud computing. While AI feels revolutionary, the security risks are not new. Threats like data poisoning and prompt injection are modern versions of classic cybersecurity problems. The danger is that AI's human-like interface makes it easy to anthropomorphize, causing users to overlook fundamental security flaws.To manage these risks, Peter advises companies to treat AI like any other software. This means applying the same rigorous security controls, testing protocols, and incident response playbooks. Instead of creating a separate process for AI, organizations should find the gaps in their current security posture and update them. This practical approach helps businesses secure AI systems effectively.Send us Fan MailSupport the show