Sixty per cent of organisations in Singapore have released software containing untested code, up from 47 per cent in 2025, according to the 2026 Quality Transformation Report from Tricentis. The sharp increase arrives as enterprises accelerate AI-driven software delivery — and as Singapore expands its AI assurance infrastructure, including the new AI Tester Accreditation Programme launched in May 2026.
The findings suggest that delivery speed, often driven by AI code generation tools, is outpacing testing discipline. Nearly two-thirds of Singapore organisations estimate that poor software quality costs them between S$641,325 and S$6.4 million annually through outages, delays, and operational disruption.
Financial Services Most Exposed
The sector most affected is financial services: over 71 per cent of financial services organisations in Singapore have shipped untested code, the highest rate across all sectors surveyed. Thirty-six per cent of respondents cite leadership pressure to accelerate delivery as the primary driver of deploying code without full testing.
Perhaps most striking is the degree of autonomous trust being placed in AI agents: 84 per cent of respondents say they trust AI agents to make software release-readiness and delivery-impacting decisions without human oversight. This combination of AI-accelerated output and reduced testing rigour creates compounding risk, particularly in regulated industries where software failures carry significant financial and compliance consequences.
Testing Governance as a Strategic Priority
Against this backdrop, Singapore’s AI Tester Accreditation Programme — which accredits firms capable of assessing AI systems for safety and governance risks — takes on additional urgency. Tricentis’s findings suggest the challenge is not only governance of AI itself, but governance of the software AI produces.
