Trust, Power & Algorithms: The Rising Stakes of AI Governance in Advocacy
The Current Wave: State-Level AI Laws and the Transparency Imperative
In 2025, U.S. states are rushing to adopt AI policies focused on transparency, accountability, and public safety. California passed the Transparency in Frontier Artificial Intelligence Act (SB 53), requiring large AI model developers to publicly disclose safety protocols and report incidents.
This trend mirrors a broader shift: states are now viewing AI not just as a technology challenge, but a civic governance issue.
For advocacy groups, this is a turning point: AI is no longer abstract; it’s a field where policy, technical systems, and public values intersect.
Why AI Governance Should Be a Core Advocacy Focus
a) Algorithms Mediate Power
Models increasingly govern access to benefits (e.g. welfare, permits), moderation, public services, credit, and even legal adjudication. If algorithms carry decisions hidden in “black boxes,” marginalized communities risk exclusion or harm.
b) Transparency Builds Trust
When agencies or services use AI, citizens deserve to know why a decision was made. Public disclosure, auditability, and recourse processes must be standard.
c) Participation in Design
AI systems inevitably encode value choices (what counts as “fair,” what gets prioritized). If communities don’t have a seat at the design table, algorithmic systems may replicate bias or exclusion.
d) Strategic Leverage
Because AI is reshaping many sectors—public services, insurance, disaster response, digital rights—advocacy groups can use AI governance as leverage to influence adjacent policy domains.
Key Challenges in AI Governance & Advocacy
Challenge | Why It Matters | Possible Strategy |
Regulatory fragmentation | State laws differ widely—leading to patchwork compliance burdens for users, vendors, communities | Advocate for model frameworks, interstate alignment, or federal baseline |
Technical opacity & “black box” models | Complex models—especially “frontier” ones—resist easy explanation | Push for model interpretability, third-party audits, and standard reporting |
Resource imbalance | Tech firms have legal, data, engineering resources; communities often don’t | Build capacity, fund civic-tech partnerships, emphasize accessible interfaces |
Insufficient enforcement | Laws may require disclosures but lack teeth | Demand whistleblower protections, audit rights, citizen complaint mechanisms |
Scope creep / mission drift | AI governance work can be highly technical and drift away from core advocacy goals | Tie AI governance to your mission (e.g. housing, disaster recovery, public services) so it reinforces, not detracts |
A Strategic Role for Unified Public Advocacy
Here’s how UPA can stake a role in AI governance aligned with its mission:
Policy Monitoring & Rapid Response Track AI bills at state and local levels. Submit comments, convene coalitions, propose amendments that center equity, transparency, oversight.
Community Audits & “Algorithm Watch” Programs Identify local services using algorithmic decision-making (e.g. licensing, benefit allocation). Engage communities in auditing those systems for bias or failure.
Participatory Design Partnerships Work with civic tech labs, universities, and government to co-design AI tools or modules (for claims processing, public service routing) that embed fairness, explainability, and citizen feedback loops.
Capacity Building & Literacy Educate community groups, local governments, and staff on AI concepts—bias, data governance, model risk. Provide toolkits for analyzing AI systems in their own neighborhoods.
Transparency & Governance Standards Promote norms, frameworks, or “civic contracts” for algorithmic systems used in the public domain: e.g. standard disclosures, audit logs, clarity of appeals.
Cross-Sector Leverage Use your convening power across sectors—insurance, disaster recovery, public service—to press for AI governance norms to be embedded in domains where UPA works.
Example Hook: AI in Public Claims & Disaster Recovery
Imagine a jurisdiction using an AI model to prioritize which disaster damage claims are processed first. If that model deprioritizes certain ZIP codes because they were historically under-reported, it entrenches inequality. UPA could:
Request full model documentation or audit access
Advocate that decision criteria be publicly disclosed
Engage affected communities to question and test model outcomes
Propose alternate rule-based fallback systems to ensure fairness
Conclusion
AI governance is no longer a niche issue—it’s central to the legitimacy of public services, civic tech, and advocacy itself. As states rush to legislate AI, organizations like Unified Public Advocacy have an opening to lead, not follow.
Our role must be clear: ensuring that algorithms serve people, not replace them; that power becomes more distributed, not more hidden; and that transparency, accountability, and participatory design are non-negotiable in the civic systems of tomorrow.