NCSC Head Urges Vibe Coding Safeguards at RSAC: 'Intolerable Risks' Demand Secure-by-Default AI Code Generation, Provable Model Provenance, and Deterministic Guardrails

UK National Cyber Security Centre·high signal

UK NCSC CEO published guidance on March 24 at RSAC warning that AI-generated code currently poses 'intolerable risks' for organizations. The four-pillar framework demands secure-by-default code generation in AI tools, provable model provenance to prevent backdoors, AI-powered code reviews on all output (human and AI), and deterministic guardrails limiting what code can do even if compromised. This is the first national cybersecurity agency to publish specific vibe coding governance recommendations.