Description of problem:

When a user enters a prompt for the first time in AI Rolling Demo Lightspeed they sometimes get a warning "No Violation". This is not blocking as the user later receives the response from the LLM.

Prerequisites (if any, like setup, operators/versions):

Steps to Reproduce

• Login to the rolling demo
• Send a prompt to lightspeed
• You'll be able to see the "No violation" warning

Actual results:

• The LLM first responds with a "No Violation" warning

Expected results:

• The LLM shoulnd't be sending this warning as a response.

Reproducibility (Always/Intermittent/Only Once):

• Intermittent: Not all the times, but the majority of times

Build Details:

• RHDH 1.8
• Model Llama 3.1 8b
• Lightspeed Plugin Version: 1.45.3 - 1.1.1

Additional info (Such as Logs, Screenshots, etc):

*** You can find the related logs (after you have reproduced) in the RHDH AI Dev Cluster > rolling-demo-ns > rolling-demo-backstage (deployment) > llama-stack / lightspeed-stack (containers)