Skip to main content
OpenIRB

OpenIRB · reviewable intelligence · open framework

Open framework for reviewable intelligence.

OpenIRB helps institutions review AI systems, autonomous agents, medical technologies, data uses, research protocols, and other high-impact decisions before they scale.

Join early accessView framework

Evidence packets · consent and authority · risk-benefit-impact · decision artifacts · continuing review

review.lifecycle
Proposal → classification → evidence → review → decision → receipt → continuing review.

Framework in formation · public standards companion at IRB.Institute · optional custody layer through EXOCHAIN

Why

Autonomous systems can act faster than institutions can review their authority, evidence, consent, and impact.

Identity proves who an actor is. Access control limits what a system can call. Logs record what happened after the fact. Governance policies describe what should happen.

None of these, alone or together, are enough to answer the institutional question:

Should this intelligent system be allowed to act, recommend, decide, learn, or scale under these conditions?

OpenIRB gives institutions a structured way to ask that question and preserve the answer.

What OpenIRB reviews

Review the system before it becomes infrastructure.

AI models

Intended use, training data, evaluation, bias, robustness, explainability, limitations, human oversight, and monitoring.

Autonomous agents

Tool access, delegated authority, scope, credentials, revocation, execution boundaries, and trust receipts.

Medical and health technologies

Clinical AI, patient data, decision support, medical workflows, safety, transparency, and post-deployment monitoring.

Research protocols

Human-subject implications, data use, consent, participant protections, AI-assisted analysis, and institutional accountability.

Data uses

Provenance, permission, minimization, privacy, vulnerable populations, secondary use, and revocation.

High-impact decisions

Civic systems, education, employment, benefits, finance, identity, access, enforcement, and organizational governance.

The OpenIRB model

Intake. Evidence. Review. Decision. Receipt. Monitoring.

  1. Intake

    Define the system, sponsor, intended use, affected groups, deployment context, and requested decision.

  2. Classification

    Identify domain, autonomy level, human impact, data sensitivity, regulatory relevance, and review path.

  3. Evidence packet

    Assemble claims, evaluations, data provenance, risk analysis, safeguards, consent posture, and monitoring plan.

  4. Review

    Human reviewers and AI-assisted tools examine evidence, risks, benefits, alternatives, uncertainty, and governance controls.

  5. Decision artifact

    Record approval, conditional approval, required modifications, denial, escalation, remediation, or continuing review obligations.

  6. Trust receipt

    Optionally preserve identity, authority, consent, policy, decision, timestamp, and custody metadata through EXOCHAIN or compatible infrastructure.

  7. Continuing review

    Monitor drift, incidents, adverse events, complaints, evidence changes, model updates, scope creep, and revocation triggers.
See the lifecycle

Artifacts, not opinions

A review is not a vibe. It is a record.

OpenIRB defines review artifacts that can be stored, audited, compared, versioned, and improved over time.

  • Review intake
  • Risk classification
  • Evidence packet
  • Consent and authority assessment
  • Reviewer questions
  • Decision artifact
  • Required modifications
  • Monitoring plan
  • Continuing review schedule
  • Incident and adverse-event log
  • Audit packet
  • Optional trust receipt

Open by design

Open where possible. Permissioned where necessary.

OpenIRB supports public-good standards and open implementation patterns while respecting privacy, security, regulated data, institutional confidentiality, and human-subject protections.

Open standards

Shared vocabulary, review models, and artifact schemas.

Permissioned evidence

Sensitive evidence packets can remain controlled while metadata and decisions are managed responsibly.

Auditable receipts

Decisions can preserve chain-of-custody without exposing protected or confidential material.

Interoperable reviews

Align review artifacts across AI governance, research ethics, quality systems, and lifecycle controls.

Ecosystem

Built to connect with the full reviewable intelligence ecosystem.

IRB.Institute

Standards, education, research, and institutional convening for Intelligent Review Boards.
Visit IRB.Institute

EXOCHAIN

Chain-of-custody for autonomous execution, authority, consent, and trust receipts.
Explore EXOCHAIN

CyberMedica

Healthcare and healing vertical for clinical AI, patient data, medical research, and whole-person care.
Explore CyberMedica

AI-SDLC Institute

AI lifecycle governance, SOPs, certification, training, and execution discipline.
Explore AI-SDLC

Make intelligence reviewable before it becomes power.

Join the early institutions, reviewers, builders, clinicians, researchers, and governance leaders shaping OpenIRB.

Join early accessRequest institutional briefing