AI coding assistant adoption reaches 97 % but governance lags at 30 % (infosecurity-magazine.com)

• 97 % of dev teams use AI coding assistants; only 30 % have full governance.
• GitHub Copilot (83 %) and Claude Code (63 %) are most used.
• 90 % of teams encounter problems with AI-generated code, mainly code review and security testing.
• Fully governed teams achieve 90 % major efficiency gains vs 44 % for non-governed.
• 64 % fear security defects; 86 % want automated vetting but 84 % insist on human review.

"A UserEvidence survey for Black Duck of 831 software engineers found 97 % actively using AI coding assistants, yet only 30 % have full governance. GitHub Copilot (83 %) and Claude Code (63 %) lead. 92 % report faster releases, saving eight hours weekly. However, 90 % encounter problems with AI-generated code, mostly code review and security testing. Fully governed teams report 90 % major efficiency gains vs 44 % for non-governed. 64 % are concerned about security defects, and 86 % want AI to vet AI code, but 84 % insist on human oversight."