arXivXiaobo Chen, Genliang Zhu, Ziyuan Wang, Chu WangThu, Nov 5, 2026, 8:41 AM PST
score 16.9
AI Code Repairs Pass Tests but Return Wrong Results
Original: Accepted but Wrong: Silent Failures of LLM Code Repairs after API Migrations
Source: arxiv.org ↗
Writing ELI5 summary…