Episode 1108
Mark Suman, co-founder of Maple AI and OpenSecret, shares his insights on how to build cutting-edge AI without sacrificing user privacy.
From secure enclaves and attestation to real-world use cases in law and finance, Mark outlines the technical and ethical foundations of private AI, and why efficient inference, not just open models, is the next big frontier.
IN THIS EPISODE YOU’LL LEARN:
00:00:00 - Intro
00:01:57 - How Mark’s Time at Apple Shaped His Vision for Secure, User-First AI
00:06:06 - Why Verifiable AI Matters for Protecting User Data
00:07:38 - The Privacy Threats of Centralized AI Models
00:15:26 - What Maple AI Does That Other AI Tools Don’t—End-to-End Encrypted, Verifiable Privacy
00:17:30 - The Threat Models Maple Addresses and How Enclaves + Attestation Work
00:19:51 - Why Inference Speed and Efficiency—Not Open Weights—Are the New AI Battleground
00:24:05 - Where Decentralized AI Fits Into Today’s Landscape
00:25:12 - A Step-by-Step Guide to Getting Started with Maple
00:29:13 - How Users Change Behavior When They Trust the AI System
00:32:42 - The Risks and Critiques of TEEs—and How Maple Answers Them
00:37:35 - Which Professions Benefit Most from Private AI
00:45:27 - Mark’s Vision for Verifiable, Private AI Over the Next Decade
Disclaimer: Slight discrepancies in the timestamps may occur due to podcast platform differences.
BOOKS AND RESOURCES
NEW TO THE SHOW?
Published on 5 days, 11 hours ago
If you like Podbriefly.com, please consider donating to support the ongoing development.
Donate