Podcast
Scott Hanselman and Mark Russinovich on AI, engineering craft, cloud, career growth, and whatever technical rabbit hole is worth following this week. New episodes every other Wednesday.
Fool’s Gold
Scott and Mark dig into the security risks of increasingly capable open-weight AI models and Mark’s research into Fool’s Gold, a defense that trains models to give convincing but deliberately flawed answers once their safety guardrails are stripped. They discuss how the technique works, whether it affects legitimate model behavior, and what it means for AI safety as open models advance.
Recent episodes
Enjoy the Journey
Scott and Mark celebrate 30 years of Sysinternals and debate when to let AI solve a problem versus figuring it out yourself.
Game
Scott and Mark compare the games they have played for years and what game design teaches about software.
Rethink Authorship
Scott and Mark explore how AI is changing software authorship, from research assistance to production code and academic writing.
Process Explorer for Mac
Scott and Mark explore Mark’s AI-assisted project to bring the Sysinternals Process Explorer experience to macOS.
Polypost: A Multi-Platform Post Editor
Scott and Mark dig into Mark’s AI-assisted Polypost project and where human judgment still does the heavy lifting.
The Cutting Room Floor
Innovations that did not make it into Mark’s Microsoft Build keynote, from AI-powered network agents to advanced datacenter cooling.
More episodes: YouTube playlist | Acast | Apple Podcasts