AI Voice Security · AI Automation Society

AI Voice Security

Hi guys,

Hope everyone’s journeys going well with your respective AI journeys!

I’m interested in people’s experiences with their agent security when deployed (or before deployment in testing).

Are you aware of any attacks taking place on active agents?
Any client info leaks?
Any downtime due to attacks?
Longer calls due to multi turn attacks resulting in higher billing for your clients?
Encoding attacks?
Multilingual attacks?
Deepfake attacks (Finance, Healthcare, Government Sector & Legal are most prone to these)
Have you tested building out agents on different platforms to see how the answers to the above questions may change due to the underlying LLM model?

Appreciate any help given here, look forward to the responses!

2 comments