42% of Committed Code Is AI-Generated. Only 48% Gets Reviewed. Here's Why That Gap Will Break Your Production.

This week, a developer ran terraform destroy and lost 2.5 years of production data. Claude Code did it. The agent was just following instructions. The story went viral. But the more uncomfortable story is in the data sitting underneath it. The Verification Debt Problem Is Real and Widening Sonar published research this week that should be required reading for any team shipping AI-assisted code: 42% of all committed code is now AI-generated Only 48% of developers always review AI-assisted code before committing 38% say reviewing AI code takes more effort than reviewing human-written code 96% of developers don't fully trust that AI-generated code is functionally correct — yet they're still shipping it Lars Janssen coined "verification debt" to describe the gap between how fast AI generates code and how fast humans can validate it. That gap is structural and it's widening. By 2027, projections put AI-generated code at 65% of all committed code. The review bandwidth isn't scaling at the sa