Large language models are open to giving incorrect, biassed or harmful responses. You should be aware of that and ensure that you are checking that the response match your expectations.

Looper isn’t designed to be a fully automated tool but rather for a person to start the process and then review the results.

Looper hasn’t been designed with any restrictions on what you can put in the prompt or the files that you’re uploading. It defers to Claude and OpenAI - and the various safety checks they have (e.g Reinforcement learning with human feedback or RLHF)