Vibe Coding: LLM-Generated Code Gains Popularity, Faces Criticism Over Vulnerabilities

What is Vibe Coding?

Vibe coding is a software development approach where large language models (LLMs) generate some or all of a project's code. AI researcher Andrej Karpathy introduced the term in February 2025, noting the increasing capability of LLMs in code generation.

Growing Popularity and Backlash

The use of vibe coding has increased as LLMs have improved. Proponents suggest it democratizes software development, enabling more individuals to create applications. However, this rise in popularity has also led to significant criticism.

Criticisms of Vibe Coding

Critics argue that vibe coding can produce vulnerable software. They also contend that extensive use of LLM-generated code can make a codebase harder to maintain, especially if the developer lacks the expertise to fix errors that an LLM cannot resolve.

Preliminary research supports these concerns. In April, Georgia Tech researchers identified 74 vulnerabilities directly attributable to AI-generated code among 43,000 security advisories over a three-month period. Fourteen of these were deemed critical security issues. Researchers estimate the actual number of AI-generated vulnerabilities is likely five to ten times higher due to disclosure limitations.

Professional Adoption

While initially popular among hobbyists, there is increasing evidence that professional developers are also adopting vibe coding, despite reservations from themselves and their peers regarding its potential drawbacks.