The Collective Intelligence Trust Stack: A Reference Architecture
A seven-layer reference architecture for evaluating whether an AI or agent system is ready for real-world deployment.
Gunjan Sinha · Trust and Safety Institute · 2025Download PDF
Community · AI Debrief Room
The AI Debrief Room
A curated TSI space for enterprise leaders, advisors, researchers, practitioners, and AI solution providers. Insights, research, recordings, and TSI Market Intelligence on what it takes to deploy AI agents with confidence.
You’re a member of the AI Debrief Room. New TSI Market Briefs arrive in your inbox as they publish.
Featured Insights
Start here.
Three pieces that set out how TSI thinks about trustworthy agents.
AI + BI = CI: Why Human Judgment Is the Missing Layer in AI Governance
Argues that Brain Intelligence (human judgment, ethics, and accountability) is the variable that turns AI into Collective Intelligence.
Gunjan Sinha · Trust and Safety Institute · 2025Download PDF
Why Trust and Safety Must Be Built In, Not Bolted On
Makes the case for building trust and safety into agent architecture from day one, so governance supports deployment instead of trailing it.
Gunjan Sinha · Trust and Safety Institute · 2025Download PDF
TSI Market Intelligence
TSI Market Briefs
Short, downloadable briefs on buyer priorities, deployment patterns, and what is working in enterprise agent programmes.
The first TSI Market Briefs are in preparation.
Members receive each brief as a download when it is published.
Articles
Articles from TSI
AI + BI = CI: Why Human Judgment Is the Missing Layer in AI Governance
Argues that Brain Intelligence (human judgment, ethics, and accountability) is the variable that turns AI into Collective Intelligence.
Gunjan Sinha · Trust and Safety Institute · 2025Download PDF
Why Trust and Safety Must Be Built In, Not Bolted On
Makes the case for building trust and safety into agent architecture from day one, so governance supports deployment instead of trailing it.
Gunjan Sinha · Trust and Safety Institute · 2025Download PDF
The Collective Intelligence Trust Stack: A Reference Architecture
A seven-layer reference architecture for evaluating whether an AI or agent system is ready for real-world deployment.
Gunjan Sinha · Trust and Safety Institute · 2025Download PDF
Research Reports
Research reports
Peer-level research on runtime assurance and governance engineering.
AI Governance Control Stack for Operational Stability: Achieving Hardened Governance in AI Systems
Proposes a layered control stack (version governance, evidence-based verification, explainability logging, telemetry, drift detection, and escalation) that treats stability as reproducible, accountable behaviour.
Horatio Morgan · Morgan Signing House · March 2026Download PDF
Continuous Governance Engineering: An Event-Driven Architecture for Evidence-Centric Runtime AI Assurance
Reframes AI governance as a continuously executing capability and introduces the Continuous Governance Operating System (CGOS) reference architecture.
Horatio Morgan · Morgan Signing House · 2026Download PDF
Advisor Perspectives
Advisor perspectives
Viewpoints from TSI’s international working group of advisors.
Advisor perspectives publish here first.
Meet the working group →Enterprise leaders, engineers, security and risk leaders, researchers, policy advisors, and solution providers share how they approach trustworthy deployment.
Industry Trends
Industry trends
Global standards and the ideas shaping how agents earn trust.
What is ISO 42001? Simple Explanation With Examples
A plain-language walkthrough of ISO/IEC 42001, the AI management-system standard: who publishes it, who it applies to, and the principles it codifies.
Dejan Kosutic · AI Governance Tutorials · November 2025Watch
Introduction to the NIST AI Risk Management Framework (AI RMF 1.0)
NIST’s own explainer for the framework, and why trustworthiness is designed in from the start.
NIST · January 2023Watch
How the MIT Media Lab Is Building a Web of AI Agents: Project NANDA Explained
Why agents need verifiable identity, and what a trust layer looks like for a world of billions of agents.
MIT Media Lab · Project NANDA · July 2025Watch
Webinars & Event Recordings
Webinars and event recordings
NIST ITL AI Webinar Series: Building Measurement Probes into Agentic AI Ecosystems
How to instrument agentic AI ecosystems for runtime measurement, the groundwork for governance and continuous learning once agents act in the world.
NIST Information Technology Laboratory · April 2026Watch
DEF CON 32: Taming the Beast, Inside the Llama 3 Red Team Process
A frontier lab’s own team on why red-teaming exists and how testing before release changes what ships.
Meta AI Red Team · Grattafiori, Evtimov, Bitton · August 2024Watch
Stanford Seminar: How Can Privacy Exist in a Data-Driven World?
How privacy can thrive alongside systems that depend on data.
Stanford University · Blase Ur, University of Chicago · May 2024Watch
Human-Centered Explainable AI (XAI): From Algorithms to User Experiences
Why an explanation only counts if the person receiving it can actually use it.
Vera Liao · Microsoft Research Montréal · February 2023Watch
Upcoming eventAI Gov World + InfoGov WorldOctober 12–14, 2026 · Las Vegas, NV
View event Discussion Topics
Discussion topics
The seven Trust Stack questions anchor every Debrief Room conversation.
- Human purposeWhat human benefit is this system designed to advance?Contribute a perspective
- Identity and authorityWho is acting, and under whose authority?Contribute a perspective
- Data stewardship and privacyCan teams collaborate without exposing data?Contribute a perspective
- Explainable executionCan a reviewer reconstruct exactly what happened?Contribute a perspective
- Simulated intelligenceWas it tested against adversarial conditions first?Contribute a perspective
- Runtime governance and incident learningCan it pause, escalate, and learn as it operates?Contribute a perspective
- Ecosystem accountabilityAre standards and lessons shared across the field?Contribute a perspective
Downloadable Resources
Downloadable resources
Papers and TSI Market Intelligence you can take back to your team.
- ArticleThe Collective Intelligence Trust Stack: A Reference ArchitectureGunjan Sinha · Trust and Safety Institute · 2025Download PDF
- ArticleAI + BI = CI: Why Human Judgment Is the Missing Layer in AI GovernanceGunjan Sinha · Trust and Safety Institute · 2025Download PDF
- ArticleWhy Trust and Safety Must Be Built In, Not Bolted OnGunjan Sinha · Trust and Safety Institute · 2025Download PDF
- Research reportAI Governance Control Stack for Operational Stability: Achieving Hardened Governance in AI SystemsHoratio Morgan · Morgan Signing House · March 2026Download PDF
- Research reportContinuous Governance Engineering: An Event-Driven Architecture for Evidence-Centric Runtime AI AssuranceHoratio Morgan · Morgan Signing House · 2026Download PDF
- TSI Market IntelligenceTSI Market BriefsDelivered to members as each brief publishes