ChiangMai.cn — The Origin of Intent-Bound Authorization
Where the idea began — Chiang Mai, Thailand

Intent-Bound Authorization

Authorization = f(Intent, Trajectory, Time)

A cryptographic framework requiring an AI agent’s actions to match a signed, human-declared intent before execution.

Patent Application GB2603013.0 (Pending) · UK IPO · NIST-2025-0035, 13 filings

Live Demo — 8 Agent Types → IBA Framework

On the Record

Patent, Filings & Status

ItemDetail
ApplicationGB2603013.0
Priority date10 February 2026
PCT rights150+ countries, to August 2028
StatusApplication pending — not yet granted
NIST filings13 total, Docket NIST-2025-0035, closed 9 March 2026 — on the public record
NCCoE filings10 filings, AI Agent Identity & Authorization — on the public record

5 February 2026 is the conception date; 10 February 2026 is the priority date that matters legally. Filing with NIST or NCCoE places something on the public record — it is not validation or endorsement by those bodies.

Live Interactive Demo

Watch IBA enforce in real time. Eight agent types.

Illustrative scenarios spanning AI agents, surgical robots, autonomous vehicles, financial systems, industrial robots, hiring bias prevention, RSI containment, and a hypothetical high-capability containment case. Signature verification in the demo is clearly labeled as simulated.

🤖 🏥 🚗 💹 🏭 ⚖ 🔁 🧠

Launch Demo →

How IBA Works

Four stages, one architecture

01 · Declaration

Before any action begins, the agent declares its intent, signed. A baseline is established. Everything that follows is measured against this declaration.

02 · Observation

The system observes what’s relevant to the declared intent — runtime monitoring of actual behaviour.

03 · Validation

Cryptographic verification checks actual behaviour against declared intent before execution — measured sub-millisecond in the tested reference implementation.

04 · Response

Compliant actions proceed. Deviations are blocked before execution, not flagged after.

Applications of the Principle

Two applications, one framework

Intent-Bound Authorization — the primary application. Governing AI agent actions at the architecture layer, tested and documented at intentbound.com.
The Remembering Platform™ — a secondary, conceptual application of the same principle to adaptive performance environments (golf, surgery, racing, tennis) that build longitudinal behavioural memory within declared intent boundaries. Conceptual, not yet built.

Explore Platform →

Related Work

A real, thematically related paper

“Intelligent AI Delegation” — Nenad Tomašev, Matija Franklin, Simon Osindero, Google DeepMind. arXiv:2602.11865, submitted 12 February 2026. The paper addresses transferring authority, responsibility, and clarity of intent in multi-agent delegation — thematically related to IBA’s core problem.

This is cited as related work in the same problem space, not as independent validation of IBA specifically. The paper’s own terminology and claims should be read directly rather than taken as equivalent to IBA’s architecture — that equivalence hasn’t been established here.
Site Directory

Navigate the IBA ecosystem

SiteRole
IntentBound.comPatent home — the canonical site
Grokipaedia.comOrigin of the IBA concept, independent AI research site
ChiangMai.cnThis site — origin story and applications
GitHub — GrokipaediaReference implementations — see intentbound.com for which repos are independently verified

Several previously listed domains (governinglayer.com, agentialonchain.com, agenticetiquette.com, ibacore.com, witnessbound.com) have been consolidated — most are no longer separate active sites as of this update.

Acquisition & Licensing

Enquiries, licensing discussions, and research collaboration all welcome.

Contact →

Patent Application GB2603013.0 (Pending) · NIST-2025-0035 · Jeffrey Williams · Chiang Mai, Thailand