What to Watch for in the New Identity Landscape

What to Watch for in the New Identity Landscape

As agentic AI systems evolve, the frameworks we use to secure and govern them must do the same. This section highlights critical constructs that have emerged to address the challenges of secure agent behavior, identity, and collaboration.

Model Context Protocol (MCP)

Model Context Protocol (MCP) is a standardized interface for how AI agents securely access external tools, APIs, or data sources. Instead of hardcoded integrations, MCP allows agents to discover and invoke capabilities through a policy-enforced gateway. It supports secure authentication, establishes scoped access, and logs every interaction.

Agent-to-Agent Protocols

Protocols like Google's Agent2Agent (A2A) secure collaboration between AI agents. They enable agents to authenticate each other, share structured goals, and delegate tasks, all while preserving security context and policy alignment.

Verifiable Credentials for Agents

Verifiable credentials (VCs) enable AI agents to present cryptographically signed claims about their identity, role, or affiliation. These credentials can be validated by relying parties without direct integration, enabling decentralized, tamper-evident verification.

Zero Trust for AI

Zero Trust principles (verify explicitly, assume breach, enforce least privilege) are even more critical when managing autonomous agents. AI agents should never have implicit trust, long-lived credentials, or unconstrained access. Applying Zero Trust means continuously verifying agents, enforcing scoped delegation, using ephemeral credentials, and applying runtime checks to all actions.