Why AI-generated vulnerability patches still require expert human review1password.com 3 pointswrxd28 days ago1 commentSaveHideCopy link On HNComments−jqpabc12328dLLMs produce Fix-Like Artifacts with Embedded Defects (FLAWED) 53.9% of the timeA human that produced similar results would likely be looking for a different job soon.
Comments
LLMs produce Fix-Like Artifacts with Embedded Defects (FLAWED) 53.9% of the time
A human that produced similar results would likely be looking for a different job soon.