AIRPP is an open-source initiative for a world where AI is already shaping records, evidence, advice, judgement, and institutional decision-making. It helps make AI-assisted work traceable, reviewable, and accountable.
Why AIRPP
Open standard that works across tools, models and platforms. No lock-in.
Built for accountability with verifiable provenance and tamper-evident records.
Places human judgement at the centre of trustworthy AI-assisted work.
Structured, interoperable, and future-proof by design (JSON-LD).
Who It's For
Create, review, and share AI-assisted records with confidence and clarity.
Integrate AIRPP to deliver verifiable, interoperable experiences.
Certify, build, and grow a trusted ecosystem together.
Meet compliance and governance needs with portable provenance.
How Adoption Works
Certified partners and adopted users help drive an open, interoperable future for AI-assisted records.
View the RegistryFoundational Custodian
The world is moving into uncharted territory at speed. AI is no longer a side tool used after decisions are made. It is becoming part of the machinery through which reports are drafted, evidence is organised, advice is framed, public debate is shaped, and institutional judgement is accelerated.
AIRPP exists because those decisions need a record. Not a vague declaration that AI was used, but a structured way to show what was produced, how AI contributed, what human judgement reviewed, and where responsibility sits.
The John Stuart Mill Institute serves as AIRPP’s foundational custodian, anchoring the initiative in principles that must outlast any single model, vendor, platform, commercial incentive, or political moment.
Principled Stewardship
If AI-assisted work is going to shape social, professional, and institutional decisions, then the provenance standard cannot be left solely to private platforms or short-term market pressure. AIRPP needs an independent public-interest anchor.
AIRPP is a community-led, open-source initiative. Your support funds development, tooling, and accessibility for all.
Public Registry
Certified partners and adopted users building trusted AI-record workflows with AIRPP.
Technical Specification
Frozen for implementation testing. Open, vendor-neutral, machine-readable provenance for AI-assisted records.
AIRPP (AI Record Provenance Protocol) is an open, vendor-neutral protocol that defines how to attach structured provenance metadata to AI-assisted records. It specifies who created a record, which AI systems were involved, which sources were used, what human reviews took place, and how the integrity of the record can be verified.
AIRPP is designed to be minimal, composable and interoperable. The core schema captures the provenance chain of a record without mandating a particular AI system, workflow tool, or delivery mechanism. Extensions allow domain-specific enrichment without breaking core conformance.
AIRPP v0.2.0 is a frozen public draft. The schema is stable for implementation testing. Extensions are in draft status. The registry is open for self-registration.
Ecosystem
AIRPP brings vendors and partners together around an open, vendor-neutral protocol for verifiable provenance, auditability and trust across tools and platforms.
Be discoverable in the public AIRPP registry and signal your commitment to trust.
Implement to the open standard and prove verifiable provenance of AI outputs.
Enable end-to-end traceability, easier audits, and stronger risk management.
Exchange provenance across tools, services and ecosystems without vendor lock-in.
Awarded to organisations that implement AIRPP to the required standard and pass conformance validation.
Awarded to organisations that have successfully integrated AIRPP and are actively using the protocol.
Join our pilot cohort to influence the standard, get early access, and grow with the ecosystem.
The Context
AI progress is not arriving as a single product or a neat administrative upgrade. It is spreading through the decision layer of society. Drafting, analysis, classification, risk assessment, evidence review, planning, professional judgement, and public communication are increasingly touched by systems that can generate and reshape material at a scale no human institution was designed to supervise manually.
That creates a basic problem. A decision can now be influenced by AI without the record clearly showing what the AI did, what information it relied on, whether a human meaningfully reviewed it, and who accepted responsibility for the final output.
AIRPP responds to that problem by making provenance portable. It gives AI-assisted work a structured record that can travel with the output, be checked later, and be understood by people outside the original system.
Core Question
That is the practical question AIRPP is designed to answer. It does not claim to prove that an output is true. It records how AI was involved, where human judgement entered the process, and what evidence of responsibility exists.
Why JSMI Matters
AI-assisted work should not rely on blind trust. People should be able to see how a record was made, what AI contributed, and where human review occurred.
Responsibility cannot be dissolved into a model, a vendor, or a workflow. AIRPP keeps human final responsibility visible in the record.
The protocol should not be captured by one platform, one vendor, or one political moment. Custodianship helps preserve that independence.
Foundational Role
The John Stuart Mill Institute serves as AIRPP’s foundational custodian. Its role is to provide principled anchoring for the initiative as AI-assisted work moves deeper into professional, civic, and institutional life.
This is not about operational control of every implementation. It is about ensuring that AIRPP remains rooted in public-interest principles: open scrutiny, individual responsibility, institutional accountability, and freedom from opaque technological power.
Visit JSMI WebsiteWhat AIRPP Makes Possible
AIRPP gives people and organisations a common way to declare AI involvement. It supports basic conformance today through the browser validator, and it creates the foundation for future registries, schema validation, attestations, APIs, and automated provenance capture.
Adopt the protocol. Integrate it into your tools and services. Partner with the initiative. Or support our mission to make AI-record provenance open, interoperable, and trusted.
Individuals and teams using AI-assisted records
Developers and organisations building AI tools
Consultancies, labs, and integrators
Researchers, non-profits, mission-aligned supporters
Complete the form and we'll guide you to the right resources and next steps.
Share a few details about yourself or your organisation.
Our team will follow up with next steps and resources.
Access tools, docs and community channels for your role.
Integrate, collaborate, and help shape trusted AI records.
Support
Everything you need to implement, integrate, and contribute to AIRPP.
Full AIRPP v0.2.0 public draft specification, schema reference, and conformance guide.
Step-by-step implementation guides for vendors, quick-start for developers, and integration walkthroughs.
Join the open-source community. Contribute to the standard, raise issues, and collaborate on GitHub.
Full REST API for manifest validation, implementation registration, and MCP-style provenance capture.
Answers to common questions about AIRPP, its relationship to other standards, and how to get started.
For partnership enquiries, press, or governance questions, reach the core AIRPP team directly.