Microsoft Podcast Explores Cybersecurity Lessons for AI Governance

Microsoft Research released a new episode in its AI Testing and Evaluation podcast series on July 14, 2025, featuring cybersecurity expert Professor Ciaran Martin (founder of the UK’s National Cyber Security Centre) and Tori Westerhoff from Microsoft’s AI Red Team.

The discussion centers on applying cybersecurity’s proven risk assessment, testing frameworks, and public-private standards to modern AI systems. Highlights include:

  • Ciaran Martin explains how cybersecurity’s tiered risk approach—ranging from small businesses to national infrastructure—can inform AI evaluation practices.

  • Tori Westerhoff draws parallels between AI “red teaming” and traditional network security, emphasizing a shared language for identifying both technical vulnerabilities and responsible‑AI failures.

  • The conversation highlights the value of embedding governance and testing into AI development early—mirroring how security is woven into software design from the start.

🎧 For those working in AI policy, engineering, or research, the episode offers a valuable bridge: practical insights from decades of cybersecurity experience that directly inform the safe, secure deployment of AI systems.

Read more here.

Submitted by Regan Williams on