Vibe coding has exploded in popularity among developers business owners for its ability to quickly generate MVPs or demo apps to showcase or get an idea of how their UX feels, and to (unfortunately) lay off software developers. However, it has pitfalls that are largely caused by irresponsible or careless practices.
Publishing Without Checking Output
Publishing an AI-generated app without checking its code first or testing it extensively poses a high risk of security vulnerabilities and catastrophic bugs that cause the app to operate differently from intended. The Garbage In-Garbage Out rule still applies.
Vibe Coding With No Software Engineering Experience
Generating apps with AI still requires supervision from software engineers. A software engineer is needed to properly assess the output of an AI-generated app. This practice causes unpatched security vulnerabilities, even after they become known. This is because AI models often fail to see the issues in their own code.
Using Agentic AI Without A Sandbox
The use of agentic AI without a sandbox could cause (and has caused) data loss and other issues. Agentic AI users have lost entire company databases because their AI agents deleted them. You also need to make backup copies of your working code outside of the sandbox. Another major issue caused by a lack of a sandbox is unwanted changes being made to production code.
Image by Thor Deichmann from Pixabay.