Bounded autonomy
The design principle that an autonomous system operates inside an explicitly declared envelope — permitted actions, value limits, counterparties, time windows — beyond which it must halt or escalate rather than proceed.
The position
This is the one idea that independently emerged in every serious agentic governance document of 2026, which is unusual enough to be worth noticing. Singapore's framework leads with assessing and bounding risks upfront; the financial services control matrix requires documented criteria and thresholds for human intervention and tested mechanisms for system shutdown. Convergent evolution in governance normally means the underlying constraint is real. The open question is not whether to bound autonomy but how the bound is expressed: a policy document is not a boundary, and a boundary that cannot be enforced in code is a preference.
This section is a claim, not a definition. It is argued rather than asserted, and it is the part of this page you are invited to disagree with. Corrections and counter-arguments to [email protected] are published.