Surge in AI-generated code is reshaping how engineers review it

A Sonar survey of more than 1,100 developers found AI contributed an estimated 42% of code added to shared codebases, but 96% of respondents said review practices need to change to catch AI-introduced errors. Teams are responding by scrutinizing plans before AI starts coding, deploying specialized review agents, routing risky changes to human reviewers, and requiring developers to defend AI-generated code.

Why it matters: This quantifies a shift the industry has mostly reported anecdotally: as AI writes a growing share of new code, the bottleneck moves from writing to verification, which could reshape engineering team structures and turn code review into its own major tooling category.