AI vs human code gen report: AI code creates 1.7x more issues

Analysis of 470 open-source GitHub pull requests reveals AI-generated code produces 1.7× more issues than human-written code. AI PRs show significantly higher rates of logic errors (75% more), readability problems (3× more), security vulnerabilities (2.74× more), and error handling gaps (2× more). While AI accelerates development speed, it lacks understanding of local business logic, repo-specific conventions, and often produces surface-level correctness without proper guardrails. Teams can mitigate these issues by providing better context to AI tools, enforcing policy-as-code, strengthening security defaults, and implementing AI-aware code review checklists.