AI has dramatically lowered the barrier to building software, accelerating both enterprise delivery and open-source contribution. According to GitGuardian, Public GitHub commits increased to about 1.94 billion, approximately up 43% year over year, while active developers grew by 33%. But this velocity has led to a lack of discipline. GitGuardian also reported nearly 29 million new hardcoded secrets leaked into public repositories in 2025, a 34% jump and a record high. MCP-related configurations alone exposed around 24,000 secrets, including over 2,000 valid credentials. In our earlier take on vibe coding, we highlighted the upside but warned that velocity without guardrails invites risk. AI coding tools are not the problem. Practices like spec-driven development show promise in enforcing consistency and auditability in AI coding. Then developers need to add proper governance. CISOs and IT leaders must enforce structured, security-first development to ensure …