Hallucination

Hallucination is when an LLM produces output that seems plausible but is factually incorrect — like importing a package that doesn't exist, using an API method with the wrong signature, or referencing documentation that was never written. In vibe coding, hallucinations are especially dangerous because the generated code might pass a quick glance but fail at runtime. This is why code review and testing remain critical even when AI writes your code.