Residual governance for responsible embodied AI: authority, contestability, and cognitive sovereignty
This article proposes a conceptual framework called residual governance for responsible embodied artificial intelligence, addressing how systems act under inevitable computational, physical, and human-state approximations. It introduces two linked mechanisms—a residual ledger to record action-relevant uncertainties and an authority ledger to track who converts residual evidence into intervention. The paper argues that cognitive sovereignty is preserved not by eliminating algorithmic influence, but by making authority transitions accountable, contestable, and reversible.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
AI researchers, developers, and ethicists studying embodied AI, human-AI teaming, and governance frameworks.
What the document contains
The document addresses the ethical and governance challenges of embodied AI systems that intervene in physical or social environments under approximation. It proposes residual and authority ledgers to make trade-offs and authority shifts transparent, and outlines a measurement agenda including robot manipulation, care escalation, and driver handover.
The text does not provide statutory legislation, binding regulatory enforcement mechanisms, or universal technical thresholds for stress and mental-state classification.
- Approximation in embodied AI is not just an epistemic limitation but a core governance condition that redistributes risk.
- A residual ledger records action-relevant remainders across physical, causal, human-state, latency, transfer, and energy dimensions.
- An authority ledger records who converts residual evidence into intervention, on what basis, with what uncertainty, and with what opportunity for contestation.
- Physiological signals must be treated as uncertainty-aware metadata rather than direct labels for mental states to protect user agency.
- Cognitive sovereignty is preserved by making AI influence accountable, contestable, and reversible when systems act under approximation.
· 1 Introduction: responsible action under approximation · 2 Approximation as a governance condition - 2.1 Relation to active inference, bounded rationality and responsible control - 2.2 The authority ledger - 2.3 Human-state information: metadata, not mental-state labeling - 2.4 Physical AI as a pressure test for governance - 2.5 Intent survivability and constrained autonomy · 3 Measurement agenda - 3.1 Worked example: physiological sensing as a residual ledger - 3.2 Limits, misuse risks and ethical implications · 4 Conclusion
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Fairness
2/5Non-discrimination and equitable treatment
What a model omits or simplifies can redistribute risk across the wider sociotechnical system, a problem already identified in analyses of abstraction in machine learning
Section 2
screening score · 2026-10-24
Human Dignity
4/5Respect for the inherent worth of persons
The central claim is that cognitive sovereignty is preserved not by eliminating all algorithmic influence, but by making influence accountable, contestable and reversible when AI systems act under approximation.
Abstract
screening score · 2026-10-24
Enforcement
2/5Consequences and compliance mechanisms with real force
Residual governance therefore adds a governance layer to resource-bounded action. It asks not only whether a system reduced an error, but whether the relevant residual was legitimate to act on
Section 1
screening score · 2026-10-24
Impact Assessment
4/5Required assessment of an AI system's risks/impacts before or during use
A residual ledger records action-relevant remainders across physical, causal, human-state, latency, transfer and energy dimensions. An authority ledger records who converts residual evidence into intervention
Abstract
screening score · 2026-10-24