Defining the object
An existential risk is not simply a very large risk. It is a risk whose realisation forecloses the possibility of correction — where the ordinary mechanism by which societies learn from harm and adjust does not operate, because there is no subsequent period in which to adjust.
That definition does more work than it appears to. It rules out most catastrophes, however severe, and it rules in a small number of hazards whose distinguishing feature is irreversibility rather than magnitude. Governance arrangements built around expected harm will systematically misprice them.
Why existing institutions underfit
International institutions are built to aggregate interests that are present at the table. The principal interest affected by an existential risk — that of people who do not yet exist — is structurally absent, and no amount of procedural reform inside a member-state body will produce it.
They are also built around thresholds of demonstrated harm. For a risk defined by the absence of a second attempt, a harm-triggered institution is not merely slow; it is pointed in the wrong direction.
Three candidate forms
The first is a standing assessment body with a statutory duty to report on irreversibility specifically, independent of any operational mandate — closer to a supreme audit institution than to a scientific panel. The second is a capability-based licensing regime that attaches obligations to the possession of a capacity rather than to its use. The third is a trusteeship arrangement in which a defined interest is represented by an office with standing to object.
Each has precedents in domestic law and none has been assembled at international level. The obstacle is not conceptual; it is that all three require states to accept constraint in advance of demonstrated harm, which is precisely the acceptance that existing institutions were designed to avoid demanding.
A note on sequencing
The sequencing matters more than the design. Assessment capacity can be built without treaty change and creates the record on which later obligations rest. Licensing follows assessment. Trusteeship, which is the most demanding, follows both.
Attempting them in the reverse order — the usual instinct, because trusteeship is the most satisfying to propose — produces a decade of drafting and no institution. The unglamorous first step is the one that makes the others available.
Placeholder text — the body of this note is indicative and awaits final copy.