Policy Research
Examine defined governance questions through evidence, analysis and attention to institutional context. The aim is to clarify a problem and its implications before moving towards conclusions or recommendations.
Our Areas of Work
Five complementary areas connect research and dialogue with institutional learning.
Comparative AI Governance
Compare approaches across jurisdictions and institutional settings while making their different purposes and assumptions visible. Comparison should explain both the points of connection and the limits of transferring an approach into another context.
Policy Dialogue
Create focused discussions around questions that need more than one perspective. The intended outcome is a clearer account of the issue, the evidence and the questions that remain, rather than an event alone.
Institutional Engagement
Connect research questions with the contexts in which institutions work. Engagement can help define relevant problems, test the usefulness of an analysis and identify opportunities for further collaboration.
Capacity Building
Explore the knowledge, capabilities and working practices needed to apply governance approaches. The emphasis is on helping people connect concepts to the responsibilities and situations they encounter.
Working Groups and Subcommittees
Contributors can bring evidence, examine a specific context, prepare comparative analysis or help refine a draft output. A clearly framed question, scope and intended audience provide the starting point.
Governance Authority & Regulatory Mapping
Legal authority, institutional responsibility, technical reach, enforcement, jurisdictional mapping and regulatory structures.
Explore this working groupHuman OS & Human Oversight Capacity
Contextual reasoning, human capability, cognitive capacity, readiness, education, qualification, intervention capability and meaningful human oversight.
Explore this working groupRuntime, Agentic & Dynamic Governance
Capability, authority, permissions, consequential action, state transitions, materiality, verification, escalation, revalidation and agentic governance.
Explore this working groupGovernance Audit, Evidence & Handover
Provenance, accountability, independent assurance, auditability, handovers between humans, systems and agents, evidence and governance continuity.
Explore this working groupLive Discussions
Live discussions bring specialist contributions into a shared conversation. A bi-weekly rhythm — every two weeks — is envisaged; confirmed sessions and participation details will be announced through AIGC news.
Institutional Engagement and Cooperation
Institutional dialogue and capacity building connect governance ideas to practical conditions. Conference participation, external publication contributions and cooperation with institutions can extend that learning. Specific activities will be listed when confirmed.
Global AI Governance Dialogue Series #9
AI and Cybersecurity: Our Greatest Defence – or Our Next Major Threat?
4:00 PM CET — as stated on the flyer
AI is helping organisations defend against cyberattacks. At the same time, cybercriminals are using AI to become more powerful. Who is winning?
View full flyer ↗Open meeting link ↗How AIGC Works
The workflow below explains how a contribution can develop. Council discussion can send questions back for refinement before synthesis, a decision proposal and approval. These discussions and decisions happen outside this website.
Governance Question
Define the problem and intended audience.
Working Group Input
Bring evidence and specialist expertise.
Council Discussion
Examine the reasoning and identify gaps.
Questions Back / Refinement
Return to contributors and revisit the discussion.
Synthesis
Integrate evidence, context and limitations.
Decision Proposal
Set out a proposed conclusion or next step.
Approval
Consider the proposal through AIGC’s agreed arrangements.
Publication / Follow-up
Share the output and learn from its use.
Questions Back / RefinementCouncil DiscussionQuestions Back / RefinementWorking Group Input
A governance question leads to working group input and Council discussion. Questions and refinements move back and forth between the Council and contributors. Synthesis leads to a decision proposal, approval and publication or follow-up. This describes work conducted outside the website.
Research to Impact
A practical contribution should show its evidence, the governance problem, what comparison adds and the implications its audience can reasonably draw.
Evidence
What is the contribution based on?
Governance Problem
Which question does that evidence help examine?
Comparative Insight
What changes or becomes clearer across contexts?
Practical Implication
What can the intended audience reasonably take from the analysis?
The editorial sequence moves from evidence to a defined governance problem, then to comparative insight and practical implication. The sequence organises an explanation; it does not remove the need to disclose uncertainty or limits.
Multi-Agent Handover Model
When work moves between agents, a handover is a governance check. Authority and constraints must remain understandable before the next agent acts.
Agent A
Prepare the contribution and its governance context.
Handover Check
Check authority, constraints and evidence.
Agent B
Receive and work within that context.
Handover Check
Verify that the governance information survives.
Agent C
Receive the next bounded task.
Action
Act within the authorised scope.
Final Compliance Check
Check the completed action against applicable requirements.
Agent A passes work through a handover check to Agent B, then through another handover check to Agent C. Action is followed by a final compliance check. A handover does not itself grant new authority.
What Must Survive a Handover
Continuity depends on carrying the relevant governance information with the work, rather than transferring an output alone.
Authority
Who or what authorises the next action, and within which scope?
Constraints
Which limits and conditions must continue to apply?
Provenance
Where did the input come from, and how has it changed?
Evidence
What supports the decision and remains available for verification?
Intervention Rights
Who can pause, redirect or stop the process, and at which points?
Observability ≠ Control
Humans do not need to approve every micro-action. Governance requires visibility of relevant state, authority, deviation and intervention points. Seeing what a system does does not, by itself, provide the ability or right to intervene.
- Autonomy requires observability.
- Authority requires boundaries.
- State change requires verification.
- Governance requires accountability.
Questions We Can Examine
These examples illustrate possible research questions and outputs. They are not claims of completed projects or publications.
Responsibility across AI handovers
Where does responsibility sit when AI work moves between systems and institutions?
This line of inquiry examines what needs to remain understandable when an output is passed to another system, person or organisation. Relevant questions include the authority to act, the constraints on the next action, the evidence available and the possibility of intervention.
- Possible output
- A research note and a practical set of questions for examining handovers.
- Intended audience
- Institutions, researchers and practitioners examining AI-supported workflows.
Governance across institutional contexts
What makes a governance approach usable in a different institutional context?
This question focuses on the assumptions an approach makes about law, responsibilities, capabilities and operating conditions. A comparison can distinguish what may be shared from what needs further examination or adaptation.
- Possible output
- A comparative paper describing enabling conditions and limits of transfer.
- Intended audience
- Policymakers, researchers and institutions comparing governance approaches.
From principles to institutional capability
What capabilities do institutions need to turn governance principles into practice?
This line of inquiry connects a stated principle with the people, knowledge and working practices needed to apply it. It asks how responsibilities can be made understandable and how practical experience can inform subsequent learning.
- Possible output
- A capability brief with a companion institutional dialogue.
- Intended audience
- Universities, public institutions and organisations developing governance capability.