The AI Rules Being Written Right Now Will Outlast Every Company in the Room

NewsOnScale Staff

October 29, 2026

There is a particular kind of political invisibility that doesn't look like suppression. It looks like complexity. It looks like a 200-page technical standards document, a comment period that closes on a Tuesday, a working group whose membership list is available somewhere if you know where to search. This is the environment in which AI governance is currently being decided — and it is not an accident.

Across federal agencies, international standards bodies, and industry-led coalitions, the rules that will govern how AI systems are built, deployed, audited, and challenged are being drafted right now. The people in those rooms are disproportionately representatives of the companies that stand to profit from favorable frameworks, lawyers hired to navigate whatever rules emerge, and a smaller number of academics and civil servants who are genuinely trying to get it right but are often outgunned in bandwidth and resources.

The public is not in the room. That's the story.

## What 'Governance' Actually Means at This Moment

When technologists and policymakers talk about AI governance, they're describing something more granular than headline-grabbing legislation. They mean: who gets to audit a model and under what conditions? What disclosures must accompany an automated decision that affects a person's housing, credit, or employment? When an AI system causes harm, who bears liability and through what mechanism can a victim even establish that harm occurred?

These are not abstract questions. They are the load-bearing infrastructure of how AI will function in civil society for the next decade. And the answers being written now will determine whether accountability is baked into the system or retrofitted later — which, in technology, almost always means never.

The challenge is structural. Regulatory agencies are under-resourced relative to the industry they're attempting to oversee. The technical complexity of modern AI systems creates genuine knowledge asymmetries that companies can exploit simply by producing documentation that takes months to interpret. Meanwhile, the lobbying apparatus around AI has grown substantially, with major technology firms spending aggressively to shape proposed rules before they become binding.

## Who Speaks for the Public Interest?

There are organizations trying to fill the gap — digital rights groups, academic research centers, some investigative journalists — but they operate at a significant disadvantage. They don't have standing in many of the convenings where language is being finalized. They often lack the technical staff to engage with model documentation at the level of specificity required. And the timelines are compressed in ways that favor established players who can mobilize quickly.

This isn't a conspiracy. It's a market structure problem with political consequences. The entities that are best positioned to participate in governance processes are the ones that can afford to monitor every relevant proceeding, employ people whose entire job is regulatory engagement, and build relationships with the officials who write the rules. Smaller companies, nonprofits, and individual citizens face a participation cost that is effectively prohibitive.

What gets lost when the public isn't meaningfully in the room isn't just idealism. It's specific protections. Transparency requirements get softened. Audit rights get narrowed. Liability carve-outs get inserted. Not always, and not by every actor — but consistently enough that the pattern is visible to anyone tracking the delta between proposed rules and final rules.

## What Accountability Would Actually Require

Meaningful public participation in AI governance would look like: proactive disclosure of who is advising regulatory bodies and what positions they're advocating. Longer, better-publicized comment periods with technical assistance for under-resourced participants. Mandatory representation of consumer and civil society interests in formal working groups. And perhaps most importantly, public access to the draft language being circulated before it hardens into policy.

None of this is radical. These are standard accountability mechanisms applied to a new domain. The question is whether there is sufficient political will — or public pressure — to demand them before the foundational decisions are locked in.

The window is not closed. But it is narrowing, one working group session at a time.