Acting on the AI Shift in Cyber Risk With Prevention at the Codebase Level
Article discusses how AI-accelerated development and cyber threats create a governance gap: traditional policies and periodic audits cannot ensure security controls operate continuously on software changes. It argues organizations must shift to prevention at the codebase level with continuous verification of all changes (human and agentic) against defined standards, producing evidence as controls operate. The EU Cyber Resilience Act and NIS2 Directive now require demonstration of implemented cybersecurity measures rather than documented policies. SonarSource proposes the Agent Centric Development Cycle framework: establish standards via quality gates, independently verify every change against those standards in the workflow and CI/CD, and remediate findings through governed paths with verified fixes.