A new proactive alignment approach for more secure code model

[XZ-X]

Hi,

Nice work! I appreciate your efforts on this survey. It gives me clearer picture for the roadmap in the code llm domain.

I found the current list does not contain alignment stage methods for code model security yet.
Our team recently makes some progress on code model security alignment. We propose a proactive method that elicits insecure code from a code model and applies fixes to it. Here's our preprint.
Hope our work can contribute a bit to the survey :)

Cheers,
Xiangzhe

[Geralt-Targaryen]

Thanks for sharing your work with us! We will add it in the next update.