When States Regulate AI First: How Public Advocacy Can Lead the Next Wave
The New Reality: States Are Leading AI Regulation
California recently signed SB 53, becoming the first U.S. state to enact a dedicated AI safety and transparency law—requiring public disclosure of safety protocols, incident reporting, and whistleblower protections.
Meanwhile, many states are introducing or debating their own AI governance bills. The U.S. Senate removed a federal ban on state-level AI regulation, signaling that states will continue to take the lead.
In tandem, the White House unveiled its AI Action Plan to streamline federal tech strategy and clarify how AI funding will interface with regulatory regimes.
The result? A fast-shifting patchwork of laws and expectations. Some states will demand transparency, others may favor light oversight. Public advocacy groups can’t wait for federal alignment—they must engage now.
Why Advocacy Matters More Than Ever
a) Shaping the Form of Oversight
When rules are still being written, advocacy groups can influence definitions—what counts as “high-risk AI,” what “transparency” means in practice, how many layers of review are needed.
b) Building Tools, Not Just Lobbying
Legal reforms need standards, audit platforms, civic-tech toolkits, and disclosure infrastructure. Advocacy groups that build or commission these tools will have critical leverage.
c) Local & Community Voices Must Be Embedded
Many AI rules are drafted by technocrats and industry. Advocates can ensure that communities historically harmed by algorithmic systems have a voice in rulemaking, not just comment periods.
d) Guarding Against Capture
Regulations may be influenced by corporations with deep resources. Civil society must act as counterweights, pushing for enforceable accountability, not just voluntary practices.
Key Tactics for Advocacy in the New Era
Tactic | Description | Why It Matters |
Draft model bills | Create template AI transparency or safety statutes and share them with state legislatures. | Helps smaller groups adopt strong language and avoids weak last-minute amendments. |
Community AI audit labs | Form local labs to test algorithmic systems (e.g. public services) and reveal biases. | Provides real evidence to push for change. |
Compute provider oversight | Advocate for requiring cloud/compute providers to log training & inference usage. | These providers are central chokepoints in AI development. (See “Governing Through the Cloud” research) |
Disclosure & appeals systems | Push for algorithms to publish decision logic, audit logs, and allow individuals to appeal outcomes. | Turns “black box” into contested terrain. |
Coalition-building across states | Unite organizations in multiple states to compare regulation drafts and share best practices. | Encourages consistency and reduces regulatory fragmentation. |
Public education & demand mobilization | Educate citizens about algorithmic harms—and demand transparency from local agencies. | Builds pressure and legitimacy behind advocacy proposals. |
A Lens: Compute Providers as Gatekeepers
One of the most promising and overlooked levers in AI governance is the role of compute providers (cloud services, supercomputing centers). Because they mediate the training and deployment of models, they are uniquely positioned to act as enforcers or record keepers in regulation. Research argues compute providers can serve as:
recorders (logging usage)
verifiers (enforcing compliance)
gatekeepers (blocking illicit model usage)
auditors (providing oversight)
Advocates should push for legal obligations on providers, not just on developers who build models.
Challenges & Pitfalls
Overregulation that stifles civic AI: Rules must allow room for public-interest models, open-source systems, and experimentation.
Regulatory capture: Influence by tech firms may water down oversight.
Fragmentation & conflict: Conflicting state rules may make compliance chaotic for public systems.
Technical opacity: Some AI models (especially emergent ones) are hard to interpret; transparency may still leave gaps.
Resource constraints: Civil society groups often lack the technical capacity to evaluate models; investment is needed.
Call to Action for Unified Public Advocacy
Launch a State AI Policy Observatory tracking AI bills and proposing amendments.
Build (or partner to build) a lightweight algorithmic accountability toolkit for community groups.
Pilot audits of local public service AI systems (e.g. in permitting, welfare, claims) to expose bias or opaque logic.
Engage with compute providers used by local agencies (cloud, data centers) and push for transparency logging.
Run public awareness campaigns highlighting how algorithms affect daily life (e.g. in insurance, benefits, municipal services).
Conclusion
We are entering a new chapter: not of AI promise vs. skepticism, but of rules that govern AI and society’s right to inspect those rules. When states lead, advocacy must lead equally — building the norms, tools, and public voice that ensure AI doesn’t quietly erode justice all around us.