Are insecure code completions a vulnerability?
Every code generation LLM model available will at some point suggest insecure code as a part of “code completion”. Should this behavior be considered a vulnerability? This post details a single concrete example I experienced with the bundled PyCharm “Full Line Completion” plugin with a project I am familiar with (urllib3).
Read more: https://sethmlarson.dev/are-insecure-code-completions-a-vulnerability
Don't miss what's next. Subscribe to sethmlarson.dev:
Share this email:
Add a comment: