Trust Identity Protocol defines verified AI agent identity
A New Spec for Agent Accountability
The AI Lab Intelligence Unobscured, Inc. published "TIP for AI Agents: Verified Agent Identity, Delegated Authority and Verifiable Actions on the Trust Identity Protocol" on November 17, 2026, authored by Dinesh Mendhe (Zenodo). The document lays out a framework for answering, in the publisher's own words, "which agent acted, on whose authority, what exactly it did, and whether a human was there when it mattered" (The AI Lab).
Verified Identity, Delegated Mandates
At the core of the release is the TIP Verified Agent, defined as an AI agent publicly linked to a verified person or company through a TIP Agent Record. Authority to act on someone else's behalf is granted through a TIP Agent Mandate — scoped, time-limited and revocable — rather than an open-ended credential (Zenodo). Actions themselves are made verifiable through signed requests and receipts, a mechanism intended to create an auditable trail for high-risk actions, including whether a human approved or participated.
What Version 1.3 and 1.4 Actually Add
Available documentation shows Version 1.3 introduced an interim deployment profile, signed endpoint definitions, agent logs, evidence states and a threat model covering 118 distinct attacks (The AI Lab). A separate record identifies Version 1.4 as published the same day, November 17, 2026. Related delegation work also surfaced this month on IETF and W3C mailing lists, including a WIMSE agent-mandate draft (IETF Datatracker) and public discussion threads (W3C).
No Confirmed Agent2Agent Profile
The original brief for this story claimed the release included interoperability profiles specifically for Agent2Agent and related protocols. That claim is not supported by the available documents. The cited material describes TIP profiles and signed endpoint definitions, but neither the whitepaper nor the Zenodo record identifies an Agent2Agent interoperability profile (Zenodo; The AI Lab). No court proceeding, regulator, penalty or dollar figure appears anywhere in the published material.
Why It Matters
As enterprises push agents into procurement, support and transaction workflows, the question of who is accountable when an agent acts has moved from theoretical to operational — a concern echoed in Microsoft's industry-wide AI transparency initiative announced the same week (Microsoft). TIP's identity-and-mandate model offers one concrete answer, even without confirmed cross-protocol interoperability yet.


