Securing LLM Systems Against Prompt Injection – Nvidia Technical Blog (developer.nvidia.com)

2 points by yandie 2y ago ↗ HN

1 comment

[ 3.2 ms ] story [ 14.5 ms ] thread

Who executes LLM-generated code (or non PR-ed code in general) against trusted environments/databases? I hope that's just a bad pattern introduced by langchain pattern and not the norm...