AIGN Research
Governing machine authority in high-consequence systems.
Applied research on machine authority, defensible autonomy and AI governance in critical infrastructure and other high-consequence environments.
From governance principles to operational models for AI-enabled systems that can inform, decide and act.
Research at the boundary between intelligence and authority
Artificial intelligence is moving beyond generating information. AI-enabled systems increasingly support operational decisions, interact with digital and physical infrastructure, and take actions with real-world consequences.
This raises a fundamental research question:
Under what conditions can authority be delegated to AI-enabled systems while keeping their actions bounded, controllable and accountable?
AIGN Research studies the authority, accountability and evidence structures needed to govern such systems. The work connects AI governance, operational risk, technical control mechanisms and institutional oversight, with particular attention to environments where system behaviour may affect public safety, essential services or organisational continuity.
Core research areas
Four conceptual research fields define what AIGN Research investigates, each addressing a distinct question. The critical infrastructure domains below provide application contexts for these research questions.
Machine authority and defensible autonomy
What authority may an AI-enabled system be granted in the first place? Topics include operational mandates, delegated authority, authority boundaries, human accountability anchors and evidence of authorised behaviour.
Operational AI governance
How is that authority embedded into real organisations, processes and systems once it is granted? Topics include organisational integration, decision ownership, operational workflows and how governance requirements translate into day-to-day operation.
Runtime governance and evidence
How is that authority technically enforced and evidenced while a system operates? Topics include policy enforcement, monitoring, intervention mechanisms, logging, auditability and the reconstruction of consequential system actions.
Cross-border and cross-sector governance
How does accountability hold when authority crosses institutional boundaries? Topics include AI-enabled systems where data, decisions and consequences cross organisational, sectoral or national lines, distributed accountability and shared operational authority.
Open research questions
Across these research areas, AIGN Research investigates questions that arise when AI-enabled systems move from supporting decisions to exercising delegated authority. These include:
- How can machine authority be formally specified, constrained and revoked?
- How can governance requirements be enforced during system operation rather than assessed only retrospectively?
- What evidence is sufficient to reconstruct and assess consequential machine actions?
- How can institutional accountability be maintained across interconnected AI systems and organisational boundaries?
The research approach explicitly examines underlying assumptions, operational applicability and limits of generalisation.
Critical infrastructure research domains
AIGN Research examines the governance of AI-enabled systems across critical infrastructure, essential services and other high-consequence environments.
Its research agenda addresses seven interconnected application domains in which delegated machine authority, operational resilience and human accountability are particularly consequential.
-
Energy & power systems
AI governance in electricity networks, generation assets, grid operations and energy infrastructure.
Research focus: machine authority in operational control, response boundaries, incident escalation and accountability for consequential actions.
-
Water & environmental infrastructure
Governance of AI-enabled monitoring, risk assessment and operational decision support in water and environmental systems.
Research focus: authority allocation, human intervention, reliability and evidence in time-critical situations.
-
Digital & telecommunications infrastructure
Governance of autonomous operations across telecommunications networks, cloud infrastructure and digital services.
Research focus: delegated operational authority, cross-system dependencies, runtime controls and recoverability.
-
Financial & payment infrastructure
Machine authority in financial operations, payments, automated control systems and essential financial services.
Research focus: transaction authority, operational resilience, human accountability and auditability.
-
Transport & mobility
AI governance across intelligent transport networks, rail, logistics and operational mobility systems.
Research focus: autonomous coordination, safety boundaries, intervention mechanisms and system-level accountability.
-
Industrial & cyber-physical systems
Governance of AI-enabled industrial processes, operational technology and systems capable of producing physical effects.
Research focus: bounded machine action, safety-critical controls, human override and evidence.
-
Civil preparedness & emergency response
Governance of AI-enabled monitoring, early warning, crisis response and systems operating across institutional boundaries.
Research focus: delegated authority, escalation, distributed accountability and governance under time pressure.
Research approach
AIGN Research combines conceptual research, regulatory analysis and applied governance architecture to investigate the conditions under which AI-enabled systems may exercise consequential authority. Its methodological approach encompasses:
- Conceptual analysis
Examining machine authority, delegated decision-making and human accountability.
- Scenario-based investigation
Analysing operational conditions, authority boundaries and the consequences of autonomous action.
- Governance architecture
Developing models for operational mandates, runtime controls and reconstructable evidence.
- Regulatory analysis
Examining the relationship between governance requirements, applicable regulation and operational implementation.
- Applied validation
Evaluating proposed governance models against relevant operational requirements and practical constraints, where suitable evidence and stakeholder access are available.
The objective is not merely to assess individual AI deployments, but to develop and examine governance models that can inform research and practice beyond a single organisational or technological context.
Research outputs may include working papers, governance reference architectures, technical and regulatory analyses, and practitioner-oriented publications.
Publications and prior work
AIGN Research builds on Patrick Upmann’s published work. Outputs are labelled by type below.
Working papers
- Upmann, P. (2025). AIGN OS – The Operating System for Responsible AI Governance. SSRN working paper. doi.org/10.2139/ssrn.5382603
- Upmann, P. (2025). AIGN Systemic AI Governance Stress Test. SSRN working paper. doi.org/10.2139/ssrn.5489746
- Upmann, P. (2025). AIGN OS – AI Agents: The AI Governance Stack as a New Regulatory Infrastructure. SSRN working paper. doi.org/10.2139/ssrn.5543162
Research reports
- Upmann, P. (2025). AIGN – Procurement Governance Gate. Zenodo, DOI-registered publication. zenodo.org/records/17936982
Books and concepts
- Defensible autonomy. Book: Defensible Autonomy: From AI Agent Capability to Accountable Enterprise Action (Amazon KDP). It examines the conditions under which AI-enabled systems may exercise delegated authority within defined boundaries, runtime controls and human accountability structures. Defensible autonomy is the author’s own conceptual model, not an established international technical standard.
- The authority supply chain. Concept discussed in The AI Governance Gap Brief (issue 32, August 2026). It asks how authority is granted, transferred, constrained and evidenced across interconnected systems, organisations and jurisdictions.
Research lead
Patrick Upmann · Founder, AIGN.Global · Research Lead, AIGN Research
Patrick Upmann has more than 20 years of experience in consulting, transformation, data and governance across finance, energy, automotive and retail. His research addresses machine authority, defensible autonomy and governance in high-consequence systems.
Recent engagement: Advanced Research Workshop, Yerevan
- Workshop: AI-Powered Monitoring Systems, Regional Data Sharing, and Models for Early Warning in Critical Services and Infrastructure, 14–19 September 2026, Yerevan State University, Armenia
- Organisers: CenTRiS Foundation (Armenia) and University of Calabria (Italy), with NATO SPS support under grant ARW.G9376
- Keynote, 16 September (Session 2.2, Governance and trust frameworks): „From innovation to implementation — building governance frameworks for AI in critical sectors“
- Round table co-chair, 17 September: „Toward a regional data-sharing framework for critical infrastructure“, with Prof. Ramaz Kvatadze (GRENA, Georgia)
This is a documented speaking engagement. It does not constitute NATO endorsement of, or funding for, AIGN Research.
Research & international cooperation
AIGN Research welcomes collaboration with universities, research institutions, critical infrastructure operators and public-sector organisations.
Potential cooperation includes joint research, scenario-based investigations, governance architecture development, applied validation and knowledge transfer.
Research collaborations are developed on the basis of clearly defined objectives, responsibilities, intellectual property arrangements and publication rights.
Project information and research findings are disclosed only where publication has been agreed and applicable confidentiality obligations permit disclosure.
Patrick Upmann · AIGN Research
Germany · International research cooperation
AIGN Research is an applied research initiative of AIGN.Global. The legal entity responsible for this website is named in the legal notice.