Can AI Access Security actually inspect RESPONSES generated by AI??

[K.Miyauchi709560]

Hello,

As stated in the subject,

The following documentation states that "AI Access Security licenses also inspect RESPONSES from generative AI applications."
https://docs.paloaltonetworks.com/ai-access-security/getting-started/introducing-ai-access-security#...

However, no matter how thoroughly I investigate the lab environment or review the documentation, I cannot find any such capability in AI Access Security.
As far as I know, enabling AI Access Security only makes a single dashboard available.

My assumption is that this statement may be a mix-up between AI Access Security and Prisma AIRS (AI Runtime Security).
On the AIRS side, there appears to be a feature called API Intercept that can scan/inspect responses from AI applications:
https://docs.paloaltonetworks.com/ai-runtime-security/activation-and-onboarding/ai-runtime-security-...

If anyone knows, please let me know.
Thanks, 

[nikoolayy1]

This sounds like DLP capability maybe when the SaaS DLP is licensed then you will get AI Access dashboards related for scanning responses if they have confidential data:

Enterprise DLP and AI Apps

AI access also is more to block shadow AI apps as Prisma AIRS Runtime AI firewall is to block prompt injections in the requests and so on. Keep that in mind.