What happened
OTT Cybersecurity LLC (Lyrie.ai, Dubai-based) announced acceptance into Anthropic's Cyber Verification Program (CVP) and released the Agent Trust Protocol (ATP) on May 11—an open, royalty-free cryptographic standard that enables real-time verification of AI agent identity, scope, authorized actions, delegation authority, and revocation status. ATP is slated for IETF submission and published under MIT license.
Why it matters
Enterprises deploying autonomous AI agents (agents that read email, write code, move money, sign contracts, manage infrastructure) currently have zero cryptographic visibility into whether they're communicating with the agent they expect, what that agent is authorized to do, or whether instructions have been tampered with. ATP is the first production-grade primitive addressing this accountability gap. Without it, agent-based attacks rely on model coercion and prompt injection rather than defeating firewalls—a structural security model shift that existing frameworks do not yet address.
Applicability
Critical for any organization deploying agentic AI at scale: finance, healthcare, government operations, cloud platforms, and manufacturing. CISOs and platform engineers should audit current agent deployments against ATP's five primitives (identity, scope, attestation, delegation, revocation) and evaluate ATP integration in agent orchestration platforms (LangChain, Anthropic, etc.). Developers should monitor IETF standardization timeline.