Any guides or best practices on how to build security layers on top of LLM agent?

API
1

Any comments on

  • Looking for keywords to prevent some cases?
  • Finding the balance of Gen AI vs program code based logic
  • Handling invisible Unicode characters?
2

Filter the output.

3

You can also take a look at the safety best practices from the docs:

https://platform.openai.com/docs/guides/safety-best-practices

It provides a good starting point.

1 Like