Establishing the Standard for the Duty of Care in AI
Moving beyond abstract ethics into verifiable compliance. AIDuty.org defines the technical and governance requirements for responsible artificial intelligence.
ADS-2026: From Principles to Verifiable Requirements
The framework moves beyond high-level ethics into technical specifications that can be audited and verified.
AI Agent Transparency & Intent Disclosure
Verification Method
Protocol-level metadata headers for non-human interactions. All automated responses must carry a cryptographic signature verifying agent identity.
Compliance Threshold
100% Disclosure compliance on public API endpoints.
GET /api/v1/inference HTTP/1.1
Host: provider.ai
X-ADS-Agent-ID: agent_77c2...
X-ADS-Signature: sig_v1_f8a2...
// Verification status: PASSED
The AI Duty Shield
Our framework is built on a core commitment to Human Wellbeing, supported by three non-negotiable pillars of technical and operational duty.
- Human WellbeingThe foundational core of all AI development and deployment.
- TransparencyVerifiable logs and explainable decision-making processes.
- NeutralityMitigation of bias and preservation of pluralistic values.
- SafetyRobust circuit breakers and fail-safe implementation.
The Three Pillars of Duty
The ADS-2026 Framework translates high-level principles into specific, verifiable requirements for AI developers and operators.
Transparency
Mandatory disclosure of training data provenance, model limitations, and real-time inference logs for high-stakes decisions.
Neutrality
Ensuring AI systems do not amplify systemic biases or manipulate public discourse through algorithmic preference.
Safety
Implementation of hardware-level circuit breakers and autonomous fail-safes for critical infrastructure AI.
Shape the Future of AI Compliance
We are seeking academics, founders, and policy experts to join our initial working group for the finalization of the ADS-2026 standard.
Apply to Working GroupPilot Case #001: AI4IP.io Implementation Report
Testing the ADS-2026 Framework against a live SaaS platform for intellectual property management. Our first pilot study evaluates real-world compliance costs and technical feasibility.