Artificial intelligence has become a standard part of modern software development. Whether developers are using GitHub Copilot, ChatGPT, Claude, GitLab Duo, or some combination of tools, AI-generated code is finding its way into codebases at an astonishing pace.
According to a new report from GitLab, however, many organizations may be embracing AI faster than they can effectively control it.
The company’s AI Accountability Report, conducted by The Harris Poll, surveyed 1,528 developers and technology buyers across six countries. The findings suggest that while AI coding tools are delivering measurable productivity gains, they are also creating new challenges around visibility, governance, and long-term maintainability.
The benefits are clear. Sixty percent of respondents said the return on investment from AI coding tools has exceeded expectations. Seventy-eight percent reported that developers are writing and committing code faster, while 73 percent said overall code quality has improved since adopting AI-assisted development tools.
Those gains help explain why AI coding tools have become nearly impossible to ignore. GitLab found that 91 percent of organizations now have at least two AI coding tools in active use, with more than half using three or more.
But speed appears to be running ahead of control.
According to the survey, 80 percent of respondents said their organization adopted AI coding tools faster than it developed policies to govern them. Meanwhile, 92 percent reported some form of governance challenge related to AI-generated code.
The report highlights what GitLab describes as a growing traceability problem. Forty-three percent of respondents said they cannot reliably distinguish AI-generated code from human-written code in their own codebase. Another 40 percent cited fragmented toolchains that fail to share context, while 39 percent pointed to systems that do not adequately track the origin of code.
Only 28 percent of respondents said their software development lifecycle tools are fully integrated with shared data and workflows, suggesting many organizations still lack a complete picture of how code moves from creation to production.
One of the report’s most revealing findings exposes a gap between confidence and reality. While 87 percent of respondents said they are confident their team could determine within 24 hours whether AI-generated code contributed to a production incident, the experience of organizations that actually faced incidents tells a different story.
Among respondents whose organizations experienced a production incident during the past year, 34 percent said they could not determine whether AI-generated code was involved.
That disconnect may become increasingly important as AI-generated code continues to accumulate inside production environments. Eighty-three percent of respondents said the growing volume of AI-generated code represents a risk that organizations need to manage now, while 82 percent warned it could create a new form of technical debt that many companies are not yet prepared to handle.
The report also found that 85 percent of respondents believe AI has shifted the bottleneck from writing code to reviewing and validating it. In other words, generating software may be getting easier, but understanding, auditing, and maintaining it remains a human responsibility.
None of this suggests organizations are planning to slow down their use of AI. If anything, adoption is likely to accelerate. What GitLab’s research makes clear is that the conversation is beginning to change. The industry has largely accepted that AI can help developers write code faster. The harder question now is whether organizations can effectively govern all of that code once it reaches production.
AI may have solved part of the productivity problem. The next challenge could be making sure developers actually know what they are shipping.
Support independent tech journalism
NERDS.xyz is independently owned and operated. If you enjoy my coverage of Linux, AI, hardware, cybersecurity, and tech culture, consider supporting the site on Ko-fi.
Support NERDS.xyz