Who Speaks for the Public When AI Rules Are Written?

NewsOnScale Staff

October 20, 2026

There is a version of AI governance that most people imagine: deliberative, transparent, with public comment periods and elected representatives weighing in on behalf of their constituents. Then there is what is actually happening — a compressed, opaque process driven largely by the entities with the most to gain from favorable rules.

The gap between those two realities is not a minor procedural footnote. It is the central accountability story of this technological moment.

## The Window Is Open, But Closing

Governance frameworks for transformative technologies rarely get rebuilt from scratch once they calcify. The decisions being made right now — about liability, transparency requirements, data rights, and how autonomous systems are permitted to act — will likely define the regulatory terrain for decades. This is not speculation. It is the pattern we saw with social media platforms, financial algorithms, and surveillance infrastructure. By the time the public understood what had been built, the building was already occupied.

What makes the current AI governance moment distinct is the speed. Regulatory bodies that typically operate on multi-year timelines are being asked to respond to capabilities that are evolving on a monthly basis. That asymmetry is not accidental. Companies deploying these systems have every incentive to move faster than the institutions trying to study them.

## Who Is Actually in the Room

The participants shaping AI policy at the highest levels break down into a recognizable set: major technology companies and their lobbyists, national security agencies, a small number of well-funded think tanks, and representatives from other governments pursuing their own strategic interests. Missing from that list is most of civil society — consumer advocates, labor organizations, disability rights groups, environmental monitors, and the public at large.

This is not an accident of logistics. It reflects resource disparities that have always shaped who gets to participate in technical standard-setting. A company with a dedicated policy team of fifty people and a government affairs budget in the tens of millions will always have more influence over draft regulatory language than a nonprofit running on grants. That dynamic did not start with AI, but AI amplifies its consequences.

The international dimension makes this more complicated, not less. As the United States and China each push to establish their preferred frameworks as global defaults — a competition playing out in standards bodies, bilateral summits, and multilateral forums — the pressure to move fast and secure advantage further crowds out slower, more inclusive processes.

## What Accountability Requires

Accountability in AI governance does not mean slowing everything down indefinitely or assuming bad faith on the part of every actor in the process. It means establishing clear records of who drafted which provisions, who advocated for them, and what tradeoffs were explicitly considered versus quietly shelved. It means impact assessments that are independent rather than self-reported. It means public comment processes that are designed to actually incorporate input rather than absorb and neutralize it.

None of this is technically difficult. Most of it is politically inconvenient for the parties currently holding the most influence.

## The Cost of Waiting

There is a recurring argument that governance should follow innovation — that regulators should wait until a technology matures before imposing rules that might constrain its development. That argument has been made, with varying degrees of sincerity, about every major technology of the past thirty years. In most cases, waiting did not produce better governance. It produced entrenched interests, locked-in design choices, and frameworks written by the industries being regulated.

The public does not need to become AI engineers to have a legitimate stake in how these systems are governed. They need institutions capable of representing that stake effectively — and right now, those institutions are lagging badly behind the pace of decisions being made on their behalf.

The room where AI's rules are being written exists. The question is whether anyone will demand a seat in it before the meeting adjourns.