In this episode of Exploited: The Cyber Truth, host Paul Ducklin is joined by RunSafe Security Founder and CEO Joseph M. Saunders and embedded systems expert Jacob Beningo to explore how AI is changing the software development lifecycle for embedded and firmware teams.

Together, they unpack the risks and responsibilities that come with AI-generated code. While AI can accelerate development and automate tedious tasks, it can also introduce defects, expand the attack surface, and create a dangerous illusion of completeness. Unlike human engineers, AI cannot explain intent, reason about long-term system behavior, or take accountability when systems fail.

Joe and Jacob discuss how engineering teams can safely integrate AI into development workflows without sacrificing security, reliability, or accountability, especially in systems that must operate safely for years in the field.

In this episode, they explore:

• Why AI-generated code can introduce hidden vulnerabilities and complexity
• The accountability challenge: who owns the risk when AI writes the code?
• How AI output should be treated as untrusted code by default
• Why rigorous testing, validation, and security reviews still matter
• Practical ways engineering teams can use AI responsibly in embedded development

For engineers, security leaders, and product teams navigating AI adoption in embedded systems, this episode offers practical insights into how to move faster with AI without weakening trust in the systems you build.