Google Wants to Write the Rules It Has to Follow

NewsOnScale Staff

June 27, 2026

There is a long tradition in American regulatory history of the most powerful players in an industry arriving early to the rule-writing table — not to constrain themselves, but to shape the constraints so they are easier to live with than whatever an uninformed legislature might impose. Google's newly published white paper on AI governance fits comfortably within that tradition.

The document, posted to the company's official blog, calls for what it describes as a pragmatic approach to AI policy in the United States. It advocates for federal preemption of state-level AI laws, risk-tiered regulatory frameworks, and a preference for industry self-governance where possible. None of these positions are unreasonable on their face. Some are genuinely defensible. But the framing of 'pragmatism' deserves interrogation, because in Washington, pragmatism almost always means something specific: workable for the people who already have scale.

## Who Benefits From Federal Preemption?

The push for a single federal standard over a patchwork of state regulations is perhaps the most consequential ask in the white paper, even if it sounds procedurally neutral. For a company operating at Google's scale, federal preemption is not just administratively convenient — it is strategically valuable. A unified national framework is far easier for a large incumbent to navigate and influence than fifty separate regulatory environments, several of which (California being the obvious example) have shown genuine appetite for aggressive consumer protections.

Smaller competitors, startups, and civil society organizations rarely have the lobbying infrastructure to engage meaningfully at the federal level. Google does. When the biggest player argues for moving the game to the arena where it has the most influence, that's worth naming plainly.

## The Risk-Tiered Framework Question

The white paper also endorses the idea of calibrating regulatory requirements to the risk level of specific AI applications — a concept that has broad support among technologists and some policy researchers, and which is not inherently industry capture. The EU's AI Act uses a similar structure. The devil, as always, is in who gets to define the tiers and who audits compliance.

Google's paper is notably light on independent oversight mechanisms. It gestures toward accountability without specifying what that looks like in practice — who conducts audits, whether findings are public, what penalties attach to violations, or how affected communities would have standing to raise concerns. A risk-tiered framework without teeth is essentially a disclosure regime, and disclosure regimes have a mixed-to-poor track record of changing behavior among well-resourced actors.

## Timing and the Regulatory Window

This paper did not appear in a vacuum. The current federal environment is arguably the most permissive toward tech self-governance in over a decade. With the EU now quietly walking back portions of its own AI Act — as reported elsewhere this week — there is a narrowing window for global standard-setters to establish a floor. Google's intervention here is an attempt to anchor U.S. policy before that floor gets set by someone else.

That is not a conspiracy. It is ordinary corporate political strategy, and it is worth understanding as such rather than either celebrating or dismissing.

## What Accountability-Focused Policy Would Actually Require

If the U.S. government is serious about AI governance rather than AI governance theater, the markers to watch are straightforward: mandatory third-party audits with public reporting, enforcement mechanisms that don't require harmed individuals to litigate their way to relief, meaningful inclusion of labor, health, and civil rights stakeholders in the drafting process, and explicit protections that don't sunset when the political climate shifts.

Google's white paper does not deliver most of those things. That doesn't make it worthless — some of its technical framing is useful — but it does mean policymakers should read it the way they'd read any document produced by a party with a direct financial stake in the outcome.

With care, and with other sources in the room.