dHub Group

A commitment to: Stay Calm: Avoiding reactionary design in the face of rapid AI change Focus on Goals: Prioritizing community outcomes over technological hype Keep an Open Ear: Centering the voices of those most impacted by digital displacement Think High-Level Governance: Moving beyond individual tools to systemic oversight Shift the Status Quo: Transitioning from private shareholder to public stakeholder Prioritize Humanity: Starting with intentionality and well-being, not perfection.

• Safe, secure and trustworthy AI
• Social, economic, ethical, cultural, linguistic and technical implications of AI
• Protection and promotion of human rights
• Transparency, accountability, and human oversight

A Sovereign Stack that integrates AI governance tier with a cryptographically-grounded technical tier. The approach moves us away from extractive data models and toward a public stakeholder status quo, addressing the critical gap between rapid AI evolution and community-led digital governance.

The Sovereign Stack lacks a sustainable foundation as long as we rely on private, extractive 'cloud' services. We argue that Cryptographically-Grounded Infrastructure (CGI) must be treated as a Public Utility, equivalent to water, electricity, or roads.

In stress-testing models against the realities of agentic AI the conclusion is clear: Trust must be embedded in the infrastructure itself, we cannot achieve trust through policy alone.

Replace extraction with stewardship by building a Sovereign Stack: a digital infrastructure where identity is owned by the individual, governance is held by the community, and AI is utilized as a tool for well-being rather than a mechanism for control. AI stewardship practice framework provides the human-centric oversight required for trust. Role: Stewardship Hub for platform governance Function: Human-in-the-loop oversight that monitors AI agent behavior against community-defined ethics and outcomes Impact: Ensures that Agentic AI remains an assistant to human decision-making, never the final arbiter.

A cryptographically-grounded utility that makes data sovereignty enforceable at the kernel level. Base: Cryptographically-Grounded Infrastructure (CGI) as a public utility Kernel: Linux Foundation Trust over IP (ToIP) ensures that every exchange - whether between humans or AI agents - is cryptographically verified. It provides the secure kernel required to prevent unauthorized data retention, ensuring non-extractive data flows Interface: While the kernel provides security, IEEE 7012 (My Terms) provides agency. This standard allows the individual to broadcast a set of machine-readable Terms of Engagement Zero-Retention Rule: The agent must prove it has flushed the data session Purpose Limitation: The agent is restricted from using data for any secondary 'training' purposes without explicit, time-bound consent Ecosystem of Ecosystems: Ayra.forum/First Person Network. This Agentic Layer facilitates the 'Onboarding' of AI agents into the social sector. By using this framework, we ensure that AI agents have a Persona that is accountable to the human community. These agents do not act on the community; they act with it, mediated by the AI stewardship human-in-the-loop oversight.

Open access

Global South Minorities

Stewardship is defined by the conditions we place on technology. Human Override: No high-stakes social outcome (e.g. mental health or resource allocation) is final without human steward validation Sovereignty Contracts: AI agents must broadcast a machine-readable Zero-Retention proof before interacting with user data Autonomous Policy Guardrails: Technical 'kill-switches' automatically freeze agentic execution if it deviates from Outcome-Based Design (OBD) parameters Community Architectural Consent: Every agentic model must be ratified by the local community D

A Sovereign Stack Framework for Public Interest AI and A Stewardship-First Approach Replace extraction with stewardship by building a Sovereign Stack: a digital infrastructure where identity is owned by the individual, governance is held by the community, and AI is utilized as a tool for well-being rather than a mechanism for control. A rigorous framework for evaluating the conditions under which agentic AI should operate in high-stakes environments and guide setting policy guardrails that ensure human-in-the-loop oversight. This moves beyond vague ethics principles toward concrete governance mandates as a structural requirement, not an afterthought. https://docs.google.com/document/d/1AbiQkwmlZF7hMXS2g9GX6yphYnfTm_ekQvFNtLZu5y4/edit?usp=drivesdk Thank you 🚀