Explore Off-by-1 Labs research on AI-generated vulnerability patches, including patch success rates, security risks, and why expert human review remains essential.