Maziyar Bahrami

Research

Research

I study how AI and digital systems become governable: how rules become operational controls, how organizations build adaptive governance capabilities, and how visibility, verification, and authority are produced across technological infrastructures.

Current Research Program

How governance becomes operational

The projects differ in setting and method, but they share a common problem: governance matters only when it can change what systems and organizations are able, required, or authorized to do.

01System level
Under reviewDesign Science ResearchHealthcare AI

Governance-by-Design of Inter-Organizational Healthcare AI Collaboration

How can governance requirements be embedded into an inter-organizational AI collaboration architecture rather than added only as external policy?

The study develops and evaluates a governance-by-design architecture for federated healthcare AI. Sensitive data remain local while participation, attribution, traceability, verification, aggregation, and accountability are coordinated through the system itself.

Theoretical focus

Governance by design, sociotechnical control, inter-organizational governance, accountability, and privacy-preserving collaboration.

Research design

Design Science Research with design requirements, artifact construction, proof-of-concept implementation, and evaluation.

Technical work

Federated learning, machine learning, blockchain-based coordination, smart contracts, privacy-preserving verification, and verification-gated aggregation.

02Organization level
Research in progressCybersecurity GovernanceCausal & Computational

From Compliance to Resilience: Cybersecurity Governance as an Adaptive Control-Loop Capability

How do organizations develop governance capabilities that can sense changing cyber risk, activate responses, monitor outcomes, and learn under regulatory pressure?

The study treats governance as an adaptive organizational capability rather than a static set of policies. It develops a theory-driven measure of governance control-loop capability from public-company disclosures and examines how strong regulatory intervention changes that capability over time.

Theoretical focus

Organizational control, dynamic and adaptive capability, regulatory response, sensing, action, learning, and cyber resilience.

Empirical setting

Longitudinal regulatory and public-company evidence, including NYDFS cybersecurity regulation and SEC annual-report disclosures.

Research design

Construct development, large-scale text analysis, panel and quasi-experimental design, causal inference, and causal machine-learning robustness analyses.

03Infrastructure level
Early-stage researchFrontier AI GovernanceInformation Infrastructure

Making Frontier AI Governable: How Interorganizational Information Infrastructures Produce Visibility, Verification, and Control Across the AI Compute Ecosystem

How do interorganizational information infrastructures make advanced AI compute governable across organizational and jurisdictional boundaries?

The project follows the governance object as it moves from physical AI accelerators to data-center and cloud compute and, increasingly, to computational activity itself. It asks how governability emerges when the resource, the information required to observe it, and the authority to intervene are distributed among different organizations.

Theoretical focus

Digital infrastructure, inter-organizational information systems, digital visibility, monitoring, verification, authority, and infrastructural governability.

Evidence

Regulatory records, organizational disclosures, infrastructure datasets, technical documentation, and a provenance-preserving governance-information event database.

Research design

Abductive, computationally augmented historical process analysis using human coding, computational extraction, temporal bracketing, negative cases, cross-process comparison, and selective network analysis.

Research Logic

Theory-led, method-fit research

I do not treat one method as the identity of the research program. The method follows the theoretical problem, the level of analysis, and the kind of evidence needed to make a credible claim.

01Define the governance problem

Identify what must become visible, controllable, accountable, or adaptable, and at what level: system, organization, or infrastructure.

02Theorize the mechanism

Use Information Systems and organization theory to specify how control, authority, information, design, incentives, or infrastructure may produce the observed outcome.

03Match method to claim

Build and evaluate an artifact, estimate a causal change, or reconstruct a process depending on what the research question requires.

04Produce usable evidence

Connect the empirical result back to theory while making clear what the evidence can and cannot establish for governance practice.

Methods by Research Problem

Different questions require different forms of evidence

The common thread is rigorous theory and transparent inference. The implementation changes with the problem.

Design and technical experimentation

When the question is how governance can be built

I use Design Science Research to derive requirements, construct sociotechnical artifacts, and evaluate whether proposed mechanisms make governance more enforceable, traceable, or verifiable.

  • Design Science Research
  • Federated learning and machine learning
  • Blockchain and smart contracts
  • Privacy-preserving verification
  • Computational modeling and simulation
  • Artifact and proof-of-concept evaluation

Causal and computational organizational research

When the question is whether organizations change

I use large-scale organizational and regulatory data to measure governance constructs and test how regulation, risk, or institutional change affects organizational capability.

  • Theory-driven construct measurement
  • Text-as-data, NLP, and machine learning
  • Longitudinal and panel data
  • Quasi-experimental and causal inference
  • Causal machine learning
  • Large-scale data processing in Python and R

Historical, process, and case research

When the question is how governance infrastructure emerges

I use heterogeneous archival evidence to reconstruct sequences, turning points, actor relationships, and changes in what can be observed, verified, or acted upon.

  • Case study and historical process analysis
  • Abductive theory development
  • Policy and regulatory analysis
  • Qualitative coding and negative-case analysis
  • Temporal bracketing and process comparison
  • Event databases and selective network analysis

Theory and interpretation

When the contribution depends on explanation

Across methods, I place strong emphasis on the theoretical mechanism and on distinguishing what the evidence shows from what remains an inference or design proposition.

  • Information Systems theorizing
  • Organization and control theory
  • Sociotechnical systems thinking
  • Digital infrastructure and inter-organizational governance
  • AI assurance, accountability, and verification
  • Comparative and mechanism-based reasoning

Broader Direction

AI governance, safety, and computational social science

My current projects define the core of the agenda, but they also open a wider set of questions about how advanced AI is governed and how computational evidence can improve organizational and policy research.

AI Governance and Safety

I am interested in organizational and frontier AI governance where control depends on concrete mechanisms: risk thresholds, evaluation and assurance, decision rights, technical safeguards, auditability, compute governance, and the infrastructures that make compliance or intervention possible.

Recent frontier AI essay →

Computational Social Science

I am interested in using large-scale regulatory, organizational, and digital data to study governance, institutions, technological change, and collective behavior. Relevant approaches include text-as-data, NLP, network analysis, longitudinal event data, causal inference, and computationally augmented case or process research.

Research Identity

Information Systems as a bridge between governance and technology

Information Systems is central to this agenda because AI governance is simultaneously organizational and technical. The field provides theories of organization, control, digital infrastructure, information visibility, design, and technological change, while also supporting empirical and design methods that can engage directly with computational systems.

This allows me to study AI governance without reducing it to either policy text or technical mechanism alone.