Developer Explains Reasons for Rejecting Functioning AI-Generated Code

*A blog post arguing against accepting AI-written code that merely works reached the Hacker News front page with 121 points.*

The post titled “When I reject AI code even if it works” appeared on vinibrasil.com and quickly drew attention on Hacker News, where it accumulated 121 points and 69 comments as of its June 21, 2026 listing.

The piece focuses on cases where code produced by AI tools passes basic functionality checks yet still gets turned down by the author. No further technical details or specific rejection criteria appear in the available source material.

Hacker News users linked to both the original article and the discussion thread. The post’s visibility on the site’s front page indicates interest among readers who work with AI coding assistants.

Why it matters

For engineers who integrate AI suggestions into their workflow, the post’s premise highlights an ongoing tension between output that runs and output that meets unstated standards of maintainability or intent. Without the full text, the exact standards remain unknown, leaving the discussion on Hacker News as the primary public record of reactions.

---

Sources:

• Hacker News (front page)

{
  "excerpt": "A blog post on rejecting AI code that works anyway hit the Hacker News front page with 121 points.",
  "suggestedSection": "software",
  "suggestedTags": ["ai-coding", "developer-practices"],
  "imagePrompt": "An abstract workspace with scattered code printouts and a single hand holding a red pen above one sheet, soft shadows on paper and wood, muted color palette, cinematic lighting, 16:9."
}