Security is sum of all parts. However vendors take an approach of Siloed and Reactive approach towards Cybersecurity. This approach also lacks Context. Moreover distributed nature of Generative AI adoption in real world is fraught with numerous issues such as Copyright Infringement, Data protection , Data Poisoning , Prompt Injection etc.
In this presentation we look at building Cryptographic Lineage and how this can be used to create an Attested Claims with Chain of Custody. This approaches starts with point to point zero trust using SPIFFE and on top of it builds Attested Claims to incorporate Context of every step of transaction. This chain of custody incorporates Zero Trust at every point of communication and adds Traceability, Audit-ability and Watermark-ability with deep rooted Trust at every step of the way. Our work is based on open-source effort of Attested Claims. We are delivering this as Zero Trust Proxy for AI which solves AI TrisM ( Trust and Security Risk Management)