AI coding agents are actually safer than human programmers at building new software, but twice as dangerous when it comes to maintaining it.
March 31, 2026
Original Paper
Safer Builders, Risky Maintainers: A Comparative Study of Breaking Changes in Human vs Agentic PRs
arXiv · 2603.27524
The Takeaway
A study of over 7,000 code changes found that AI is remarkably good at creating new features without breaking anything. However, when asked to perform routine maintenance or 'cleanup' tasks, AI agents are significantly more likely than humans to accidentally break the entire system due to a 'confidence trap.'
From the abstract
AI coding agents are increasingly integrated into modern software engineering workflows, actively collaborating with human developers to create pull requests (PRs) in open-source repositories. Although coding agents improve developer productivity, they often generate code with more bugs and security issues than human-authored code. While human-authored PRs often break backward compatibility, leading to breaking changes, the potential for agentic PRs to introduce breaking changes remains underexp