Survey Surfaces Rise in IT Incidents Attributable to AI Coding Tools

A comprehensive survey involving 406 IT decision-makers from North American organizations with over 250 employees has brought to light a significant trend: 93% of these organizations have encountered at least one infrastructure incident directly linked to their use of artificial intelligence (AI) coding tools. This study, commissioned by Spacelift and carried out by Panterra Group, indicates that the rapid adoption of AI in software development, while intended to enhance efficiency, is concurrently introducing new complexities and risks. The survey's findings reveal that AI's influence extends beyond mere code generation, placing increased demands on infrastructure teams. Specifically, 86% of respondents reported that AI has intensified these demands, manifesting as quicker emergence of security vulnerabilities (40%), more challenging governance (40%), accelerated change rates (37%), heightened strain on CI/CD pipelines (35%), and a growing problem of infrastructure drift (35%). These statistics underscore a critical challenge: the pace of AI-generated code is outstripping the ability of existing governance policies to manage it effectively. A particularly concerning insight from the survey is the lack of rigorous oversight for AI-generated infrastructure-as-code (IaC). Only 15% of organizations track the volume of AI-generated IaC moving through their pipelines, and a mere 20% monitor the error rates of AI-generated changes. Alarmingly, one-third (33%) of infrastructure teams admitted to applying AI-generated HashiCorp Configuration Language (HCL) code directly to production without any review, while another 43% conduct only a minimal review. This lax approach significantly increases the risk of vulnerabilities that malicious actors could exploit rapidly. Dimitri Vlachos, Chief Marketing Officer for Spacelift, emphasized that AI is profoundly impacting DevOps teams by introducing code of uncertain quality into the infrastructure provisioning process. The survey categorizes respondents into four AI maturity levels: Pioneers (19%), Outpacing (25%), Fragmented (32%), and Exposed (24%). Pioneer organizations, while using AI-generated IaC at a high rate (86%), do so within governed pipelines equipped with automated validation and policy enforcement, suggesting a more secure adoption model. The report highlights the urgent need for organizations to develop robust governance policies that can keep pace with AI's capabilities. It suggests that as the volume of code flowing through DevOps pipelines continues to grow, so too will the incidence of software-related issues requiring investigation by engineers. The implication for CI/CD is clear: pipelines must evolve to incorporate more sophisticated AI-aware validation, security, and governance mechanisms to mitigate the risks introduced by AI coding tools.