Wednesday, September 17, 2025

The Digital Press

All the Bits Fit to Print

Ruby
Web Development Artificial Intelligence Urban Planning Astronomy

AI Code Detector Achieves 95% Accuracy in Identifying AI-Generated Code

Machine learning model detects AI-generated code with 95% accuracy

From Hacker News Original Article Hacker News Discussion

Span has developed an AI Code Detector that can identify AI-generated code with 95% accuracy for TypeScript and Python by analyzing semantic code patterns without needing integration with specific AI tools.

Why it matters: Provides engineering leaders clear visibility into AI-generated code presence and quality in their repositories.

The big picture: The tool helps track AI vs. human code ratios, defect rates, and plans to measure AI coding tool impact on productivity.

The stakes: Some code chunks (about 10%) are marked "unknown" due to insufficient data, and accuracy varies by code chunk size.

Commenters say: Users want more precision/recall metrics, support for additional languages, and integration into CI/CD pipelines for practical use.