RightWalk AI Policy
Date: 30 May, 2025
Purpose
To establish a comprehensive ethical, operational, and governance framework for the development, deployment, and long-term stewardship of AI tools at RightWalk, ensuring that artificial intelligence advances our mission of equity, empowerment, and effective public service delivery.
Mission and Purpose
RightWalk is dedicated to uplifting 50 million individuals from poverty into dignified and sustainable living by 2030. This mission guides the ethical framework of our AI practices, ensuring that technology serves those who need it most.
AI Use
AI tools will be utilized primarily for enhancing beneficiary services, focusing on community engagement and crisis alerting. These tools will help streamline operations and improve the effectiveness of the government’s social policies.
1. Guiding Principles
1.1 Equity First Design
Every AI tool developed or deployed by RightWalk must prioritize historically marginalized communities, reduce systemic exclusion, and work to dismantle barriers based on caste, class, gender, geography, or disability.
1.2 Transparency and Consent
AI systems must be intelligible to their users. We will ensure transparency about how data is used, how decisions are made, and what limitations exist. Users’ consent—especially for sensitive data—will be informed, recorded, and revocable.
1.3 Data Anonymization & Protection
Personally identifiable data, especially of children and vulnerable groups, will be anonymized. Data will be stored in secure environments with access controls aligned with best-in-class privacy standards.
1.4 Fairness & Bias Mitigation
RightWalk will regularly test its AI systems for bias across demographic dimensions (gender, caste, disability, etc.). Corrective measures—including retraining datasets and introducing rule-based overrides—will be implemented as needed.
1.5 Human Oversight & Accountability
AI will augment—not replace—human decision-making. All critical outputs (eligibility, resource access, predictions) will have mechanisms for human override and redressal.
1.6 Language & Accessibility
AI interfaces will be multilingual, mobile-first, and designed to meet the needs of users with low literacy or connectivity. Accessibility for persons with disabilities will be incorporated into product design.
1.7 Open Knowledge & Interoperability
Wherever possible, RightWalk will use open-source AI models, and its tools will follow open standards to enable interoperability across government and civil society systems.
2. Ethical Use & Data Governance
2.1 Purpose Limitation
Data collected or used for training AI will be limited to specified, lawful, and well-communicated purposes.
2.2 Data Ownership
RightWalk will assert ownership of internally generated data while respecting institutional agreements when working with government systems.
2.3 Informed Consent
RightWalk will ensure users know when they are interacting with AI, what data is being collected, and how it will be used.
2.4 Child and Youth Protection
AI systems interacting with children will be reviewed by domain experts in child rights and education.
2.5 Auditability
All major AI decisions (e.g. recommendations, predictions) will be auditable.
2.6 External Reviews
Periodic third-party reviews will ensure AI systems remain aligned with RightWalk’s values and Indian legal frameworks.
2.7 Community Feedback
Feedback from community surveys integrated into AI tools (like chatbots) will be regularly collected and analyzed to refine our interventions.
2.8 Data Stewardship
Dedicated data stewardship committees, including community representatives,will ensure participatory data governance and reinforce data rights of the communities served.
3. Deployment & Monitoring
3.1 Stage Gated Development
AI solutions will pass through three stages—internal validation, ethical risk assessment, and controlled field pilots—before scale-up.
3.2 Feedback Integration
All AI tools will include user feedback channels that feed directly into product improvement loops.
3.3 Real-Time Monitoring
Live dashboards and alert systems will track the health, usage, and impact of AI deployments.
3.4 Contingency & Rollback
Systems must allow for graceful fallback or shutdown in the event of failure, error, or misuse.
3.5 Bias Audits and Monitoring
AI systems will undergo real-time monitoring and regular audits for bias. Community involvement in these audits will help ensure fairness and contextual accountability.
3.6 Human-in-the-Loop Systems
Escalation protocols will be embedded to ensure that complex or sensitive decisions involve human review.
4. Collaboration and Capacity Building
4.1 Co- Development Ethos
RightWalk will co-create AI tools either in-house or with partner institutions (e.g.,IDInsight, Turn.io, Edzola), government departments, and frontline actors to ensure contextual relevance and ownership.
4.2 Public Sector Integration
AI solutions for government will align with Digital India standards and be hosted on secure infrastructure (e.g., AWS India Region, NIC Cloud).
4.3 Capacity Building
Training programs for frontline staff, data teams, and policy units will be embedded into all deployments.
4.4 Knowledge Sharing
AI playbooks and deployment guides will be published, where possible and approved, to support replication by other civil society organizations.
4.5 Third Party AI Tools
Contracts with external AI service providers will define accountability structures, and periodic reviews will ensure adherence to ethical and operational standards.
4.6 Staff Training
AI ethics and responsible use will be part of ongoing staff training through workshops and informal learning platforms.
5. Review, Evolution and Governance
5.1 Annual Policy Review
This policy will be updated annually based on legal developments, user feedback, and implementation learning.
5.2 AI Strategy Review Board
An internal governance body will monitor AI initiatives, approve new deployments, and oversee ethics compliance.
5.3 Alignment with National Frameworks
All AI efforts will align with India’s Digital Personal Data Protection Act, Responsible AI guidelines by NITI Aayog, and any sectoral regulations.
5.4 Impact Assessment
Each AI deployment will include pre- and post-assessments to measure effectiveness and inform adjustments.
5.5 Legal and Regulatory Compliance
Compliance with data privacy and AI-related laws will be maintained through regular legal reviews, and expert consultations will guide compliance efforts.