Zero-Shot Learning

The new attack surface writes code | Travis McPeak from Cursor

Episode Summary

For Travis McPeak, chaos has always been part of the job of securing companies like Netflix, Databricks, and IBM. Now, as Cursor's Head of Security, he's tackling the industry-wide problem of securing AI-powered software development. In conversation with 1Password CTO Nancy Wang and Google's Dev Tagare, Travis rejects the idea that security is a blocker and challenges the assumption that agents can be trusted users.

Episode Notes

In this episode:

• Why AI agents can't be trusted users
• The hardest security decisions are about what not to protect
• Everything in security is a proxy, including agents
• Secure-by-default doesn't mean slowing productivity
• AI-generated code isn't vibe-coding if a human verifies the work

Zero-Shot Learning is a builder-to-builder podcast about how AI systems are designed, deployed, and secured. Subscribe for more.

Go deeper:

Episode companion blog: https://www.1password.com/blog/never-trust-your-coding-agent

1Password Developer newsletter: https://1password.com/developer-newsletter

Travis:

LinkedIn: https://www.linkedin.com/in/travismcpeak/

Connect with 1Password

1Password.com

1Password Developer newsletter: https://1password.com/developer-newsletter

Build securely with 1Password Developer: https://developer.1password.com/