AGI Governance: building governance for advanced AI
agigovernance.org is an independent-facing digital observatory dedicated to the governance of Artificial General Intelligence and highly capable AI systems. Its purpose is to map emerging principles, institutional models, safety frameworks and international policy approaches for technologies whose capabilities may exceed the assumptions of traditional software governance.
AGI is not currently a standalone legal category under the EU AI Act. Existing European rules address, among other things, general-purpose AI models and impose additional obligations on models presenting systemic risk. AGI Governance treats these frameworks as part of a broader transition toward governance structures for increasingly capable and autonomous AI.
The pillars of AGI governance
Institutional governance
Advanced AI governance requires clear allocation of authority, responsibility and oversight across developers, deployers, boards, regulators and independent evaluators.
Safety, security and systemic risk
Governance must integrate model evaluation, adversarial testing, misuse prevention, infrastructure security, incident reporting and systemic-risk assessment.
Policy and regulatory readiness
Advanced AI governance is emerging through regulation, voluntary codes, technical standards and international initiatives. The project tracks frameworks relevant to frontier AI and future AGI-class systems.
Accountability and human control
Meaningful human oversight requires escalation thresholds, authorization mechanisms, access controls, intervention procedures and verifiable responsibility for consequential decisions.
Evaluation, audit and assurance
Independent testing, documentation, traceability and assurance provide the evidence layer of governance and turn principles into demonstrable controls.
International coordination
Highly capable AI systems operate across jurisdictions. Shared standards, interoperable evaluations, incident communication and international cooperation may become essential.
Building governance before the category is formalised
The core premise of AGI Governance is anticipatory: governance frameworks should mature before capabilities outpace existing controls. The project does not claim that a specific AGI regulatory regime already exists; it provides a structured place to examine the concepts, institutions and assurance mechanisms that may shape it.
Institutional reference sources
The project uses the current GPAI and systemic-risk framework as a legal reference point while keeping prospective AGI analysis distinct from existing legal categories.