Assessing security implications of using generative AI in software development in Tanzania: A quantitative study

Generative artificial intelligence (AI) is increasingly used in software development, raising concerns about the security of AI-generated code. This study assessed the security implications of generative AI use in software development in Tanzania. A quantitative, cross-sectional survey design, guided by the Technology Acceptance Model (TAM) and Protection Motivation Theory (PMT), was employed to collect structured questionnaire responses from 77 software developers. Descriptive statistics and non-parametric tests, including Spearman’s rank correlation and the Kruskal–Wallis test, were used to examine four hypotheses. The findings showed that 78% of respondents used generative AI tools daily or several times a week, with ChatGPT being the most commonly used tool (68.8%). Respondents generally perceived these tools as easy to use (mean = 3.94), while trust in the security of AI-generated code was relatively low (mean = 2.93). Programming experience was significantly associated with trust in AI-generated code security (H(4)=17.431, p=.002). Respondents with less than one year of programming experience reported higher trust than several more-experienced groups. However, no statistically significant relationships were found between AI usage frequency and perceived security risk, AI tool choice and reported security issues, or perceived ease of use and perceived security risk. The findings highlight the importance of reviewing AI-generated code, applying secure coding practices, and conducting security testing. Larger studies incorporating objective code-security assessments are recommended to strengthen the evidence.