Who Gets to Write the Rules for AI? Not You.

NewsOnScale Staff

October 27, 2026

There is a particular kind of policy urgency that benefits from public confusion. When a technical subject feels too complex for ordinary people to weigh in on, the window for expert consensus closes fast — and the experts invited to that table are rarely neutral.

That dynamic is now playing out in real time across AI governance. The decisions being made in the next 12 to 18 months — about liability, transparency requirements, algorithmic accountability, and which government bodies hold enforcement power — will shape how AI systems affect employment, civil liberties, financial access, and democratic participation for decades. And the public, by nearly every measurable indicator, is not part of the conversation.

## Closed Rooms, Lasting Consequences

Federal agencies, congressional staffers, and industry working groups are currently drafting frameworks that determine when an AI system must disclose its nature to a user, when a company can be held liable for automated decisions, and whether state-level consumer protections can coexist with federal standards. These are not abstract regulatory questions. They determine whether a denied loan, a wrongful content removal, or a biased hiring algorithm has any meaningful remedy attached to it.

Public comment periods exist in theory. In practice, they are dominated by organized industry stakeholders who have the legal and communications infrastructure to participate at scale. A form-letter campaign from concerned citizens carries negligible weight against a 200-page technical filing from a major platform's policy team. The asymmetry is not incidental — it is structural.

Civil society organizations working on AI accountability have flagged this gap repeatedly, but their warnings tend to surface in specialty publications rather than mainstream coverage. That itself is part of the problem: when governance debates stay inside the policy press, the feedback loop never reaches the people most likely to be affected by the outcomes.

## The Preemption Problem

One of the sharpest immediate conflicts involves regulatory preemption — the legal doctrine that federal law can override state-level rules. Several states, particularly California, Colorado, and Illinois, have moved ahead of Congress with AI-specific consumer protections. Federal AI legislation, if and when it passes, may nullify those protections entirely, replacing them with weaker uniform standards that were more palatable to industry negotiators.

This is not a hypothetical concern. It is the same playbook used in financial services regulation and data privacy, where federal frameworks have historically set ceilings rather than floors — limiting what states can do rather than guaranteeing minimum protections for residents.

For anyone watching how the agent economy is developing — AI systems that act autonomously on behalf of users, make purchases, send communications, negotiate contracts — the stakes of getting preemption wrong are unusually high. Once those systems are embedded in everyday economic life, retroactive accountability becomes nearly impossible.

## What Accountability-Focused Coverage Has to Do With This

NewsOnScale covers AI governance not because it is a niche regulatory beat, but because it is the infrastructure story of this decade. The rules being written now will determine whether AI agents are accountable to the people they serve or primarily to the platforms that deploy them.

That distinction matters enormously. An AI agent that manages your calendar, screens your job applications, or interfaces with your bank on your behalf is only as trustworthy as the framework that governs its behavior. If that framework was written without your participation, by people whose primary obligation is to shareholders rather than users, you should know that — and you should factor it into every assumption you make about what these systems will do when your interests and a platform's interests diverge.

The question of who was in the room when the rules were written is not a process complaint. It is a power question. And on AI governance, the answer, so far, is straightforward: not you.

We'll keep reporting on who is.