作者位次 #3/9ACL 2026Poster· Natural Language Processing
When "Correct" Is Not Safe: Can We Trust Functionally Correct Patches Generated by Code Agents?
Yibo Peng · James Song · Lei Li · Xinyu Yang · Mihai Christodorescu · Ravi Mangal · Corina S. Pasareanu · Haizhong Zheng · Beidi Chen
摘要
Code agents are increasingly trusted to autonomously fix bugs on platforms such as GitHub, yet their security evaluation focuses almost exclusively on functional correctness. In this paper, we reveal a novel type of threat to real-world code-agents: functionally correct yet vulnerable (FCV) patch...