Bringing Privacy to AI Models with TrustedRouter
Summary
Host Rich Straolino interviews Joseph Perlo, CEO of Trusted Router, alongside security experts TC Nijakowski and Keith McCartney to address data privacy risks in AI. The episode explores how Trusted Router uses confidential computing and remote attestation to ensure prompts remain encrypted and verifiable, preventing data leakage to model providers. Perlo explains how this open-source solution offers end-to-end security while reducing AI costs by up to 90% through efficient routing. The discussion also covers the limitations of zero-data-retention contracts and the importance of auditable s…
Topics discussed
Introduction and guest introductions
Why AI privacy remains a challenge
Trusted Router value proposition and architecture
Upstream provider security and the Private AI Alliance
End-to-end encryption and remote attestation details
Model coverage and jurisdiction controls
SOC 2 compliance scope and threat model
Cost savings and ease of implementation
Incident response and immutable logging
Integration with existing internal tooling
Critique of zero data retention and open source benefits
Closing remarks and show credits
Listen ad-free on Castria