Vibe coding to vibe hacking: securing software in the AI era (www.techradar.com)

“Vibe coding” — using generative AI to write large amounts of code from natural-language prompts — is accelerating software development (Gartner forecasts AI-assisted coding could power ~40% of new business software within three years). It can eliminate tedious boilerplate, speed refactoring and experimentation, and even sharpen code-review skills by letting devs focus on higher-level design. But the flip side is a widening attack surface: developers become distanced from hand-authored source, AI can introduce insecure patterns at scale, and sheer volume of generated code can raise the aggregate defect rate unless rigorous review and testing are enforced. The security implications are acute: adversaries are also using generative AI (“vibe hacking”) to craft exploits and inject malicious code, shortening the window defenders have to patch or detect attacks — attackers can weaponize known flaws faster than upstream patches propagate. The remedy is technical and procedural: provenance, dependency and supply-chain auditing, continuous integration checks, static/dynamic analysis, stronger code-review guardrails, and embedding security into the SDLC so developers act as stewards of AI output. In short, automation must be paired with accountability, transparent provenance, and continuous verification to make AI-driven productivity sustainable and secure.