Why OpenClaw Is The Most Dangerous Form Of AI Yet - Illia, Eli & Joel

OpenClaw is not as safe as you think. Over the last few weeks, many AI agents are leaking credentials and have become susceptible to prompt injections.

We need something better.

In today's episode of AI Supercycle, we host Illia Polosukhin, NEAR Co-founder, unveils IronClaw, a security-first framework isolating credentials from prompt injections using OpenClaw. Joel Valenzuela, Director of Marketing at Dash explores privacy in agentic payments, and Eli Ben-Sasson, co-founder and CEO at Starkware, discusses post-quantum security.

Together, they reveal why privacy infrastructure determines whether AI works for you or exploits you.

We cover:

- Iron Claw: Privacy-Protecting AI Framework
- How OpenClaw's Security Flaws Got Patched
- Agent-to-Agent Commerce & Marketplaces
- Why Chain Abstraction Matters for AI
- Post-Quantum Crypto Security
- HTTP 402: Internet-Native Micropayments

The Rollup