Security Weekly - A CRA Resource: AI Broke the Shift-Left Model

Shift-left security was built around a human-driven development workflow. AI coding agents can now perform much of that workflow themselves, from reading an issue to opening a pull request.

The first commit may no longer be the earliest meaningful security boundary. The critical moment could instead be when an agent receives the context and authority to act.

That creates a new question for AppSec and DevSecOps teams: what does “left” actually mean when an agent starts working before the code exists?

Where should security controls begin in an agent-driven development workflow?

Subscribe to our podcasts: https://securityweekly.com/subscribe

#AISecurity #AppSec #DevSecOps #SecurityWeekly #Cybersecurity #InformationSecurity #AI #InfoSec