LLM-generated OS code produces unreliable and buggy systems software

Developers using LLMs to generate low-level systems code like operating systems encounter hallucinations and structural bugs that make the result non-functional and hard to verify. The vibe-coding approach breaks down at the systems level where correctness guarantees matter. This experimental project exposes the limits of AI code generation outside of high-level application domains.