Generative AI tools like GitHub Copilot and ChatGPT have rapidly gained traction in software development, offering automation and efficiency. However, these tools also raise critical questions about trust and security.

This study explores how usage frequency, programming experience, tool choice, and perceived ease of use influence developers’ perceptions of security risks and trust in AI-generated code. Using a quantitative survey of 77 developers, we tested four hypotheses guide from theoretical frameworks, such as the Technology Acceptance Model and Protection Motivation Theory.

The findings reveal that programming experience significantly impacts trust in AI-generated code security. However, no statistically significant relationships were found for usage frequency, AI tool choice, or perceived ease of use and their associations with perceived security risk and encountered security issues. Descriptive insights highlight that while most developers find AI tools easy to use and widely adopt them (78% frequent users), overall perceptions of security risks remain moderate and trust in AI-generated code remain low.

Overall, these results underscore the need for rigorous security evaluations and the development of guidelines for safe AI integration in software development. Organizations should prioritize educating developers on AI limitations and implementing robust security protocols for AI-generated code.