Generative AI impacts open source development and code review processes

A Linux Foundation survey reveals that generative AI is significantly altering the open source landscape. While AI has reduced code-authoring time by an average of 52 percent, it has also led to an influx of low-quality contributions, often referred to as “AI slop.” Maintainers estimate that approximately 25 percent of incoming pull requests are low-quality, a figure that rises to 31 percent for larger projects.

Beyond code quality, the technology is impacting project governance and security. Only 35 percent of projects currently have formal policies regarding the use of generative AI, leaving many contributors and maintainers without clear expectations. Additionally, security teams are facing increased workloads due to an influx of AI-assisted reports, prompting some projects, such as curl, to adjust how they handle vulnerability disclosures.

In broader software engineering, reports indicate a significant shift in how code is reviewed. Data from Faros AI shows a 76.3 percent increase in pull requests being merged without manual review since April, as many companies now utilize AI to handle between 50 percent and 80 percent of their reviews. While AI excels at mechanical bug-finding, experts note it lacks the ability to facilitate the high-level architectural debates and decision-making processes essential for building a team's mental model of a system.