Training AI code generators to catch their own security bugs

Original: Learn from Your Mistakes: Tree-like Self-Play for Secure Code LLMs

Source: arxiv.org ↗