In this episode, Nick Lambert and Peter Horadan tackle the rising identity crisis in AI. With digital agents poised to outnumber humans online, how can we ensure they act safely and within our control? They discuss the key risks, legal gaps, and solutions like MCPI, a framework that redefines how we trust and manage AI agents.
AI agents are multiplying fast: By next year, there may be more non-human agents online than human users, automating tasks from travel booking to work management.
Identity is critical: Who is the agent acting on behalf of? What permissions do they have? How can misuse be prevented?
Five identity challenges for AI agents:
Agent Identity – Each agent needs a verifiable identifier (DID).
Delegation – Human authority must be granular and revocable.
Reputation – Agents require reputations; malicious actors must be identifiable.
Human Identity – Humans must be verified securely once.
Legal Agreements – Agents cannot accept legal terms; humans must consent.
Demo: Know Your Agent in Action – Booking a flight via Atlas AI illustrates secure identity verification, delegation, and durable authorization.
MCPI (Model Context Protocol – Identity): Adds identity, delegation, legal consent, and agent reputation to Anthropic’s MCP framework. Open standards with SaaS implementation.
Digital IDs & Wallets: EUDI wallets in the EU will enable cryptographic signing and selective disclosure of personal attributes.
Practical guidance for businesses: Block untrusted AI agents from internal systems, plan for MCP servers, and rethink identity workflows for non-human actors.
Peter Horadan — CEO, Vouched
Peter leads Vouched in transforming how companies verify identities instantly and securely, even for non-human actors. With a background in scaling automation at companies like Concur, Avalara, and Microsoft, he brings expertise in compliance, cybersecurity, and operational efficiency.