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
Patent, Filings & Status
| Item | Detail |
|---|---|
| Application | GB2603013.0 |
| Priority date | 10 February 2026 |
| PCT rights | 150+ countries, to August 2028 |
| Status | Application pending — not yet granted |
| NIST filings | 13 total, Docket NIST-2025-0035, closed 9 March 2026 — on the public record |
| NCCoE filings | 10 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.
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.
🤖 🏥 🚗 💹 🏭 ⚖ 🔁 🧠
Four stages, one architecture
Before any action begins, the agent declares its intent, signed. A baseline is established. Everything that follows is measured against this declaration.
The system observes what’s relevant to the declared intent — runtime monitoring of actual behaviour.
Cryptographic verification checks actual behaviour against declared intent before execution — measured sub-millisecond in the tested reference implementation.
Compliant actions proceed. Deviations are blocked before execution, not flagged after.
Two applications, one framework
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.
Navigate the IBA ecosystem
| Site | Role |
|---|---|
| IntentBound.com | Patent home — the canonical site |
| Grokipaedia.com | Origin of the IBA concept, independent AI research site |
| ChiangMai.cn | This site — origin story and applications |
| GitHub — Grokipaedia | Reference 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.
Enquiries, licensing discussions, and research collaboration all welcome.