Political Tech
As governments compete to shape AI rules, the absence of coordination is becoming a feature, not a bug — and the costs will fall on everyone else.
NewsOnScale Staff
July 10, 2026
There is a particular kind of political paralysis that doesn't look like paralysis. It looks like summits, white papers, interagency task forces, and carefully worded joint statements. It looks like progress while producing none. That is roughly where global AI governance stands in mid-2025 — a sprawling, overlapping tangle of national frameworks, international bodies, and industry self-assessments that clash more often than they converge.
The core problem isn't technical. Engineers can describe what large language models do with reasonable precision. The problem is jurisdictional and political: no government wants to cede meaningful authority over a technology it believes will define economic and military power for the next several decades. The result is a regulatory environment that is simultaneously overcrowded and underenforced.
## Competing Visions, Competing Interests
The European Union's AI Act represents the most comprehensive attempt to build a binding legal framework — tiered by risk level, with requirements for transparency, human oversight, and conformity assessments for high-stakes applications. It is also, by design, a market-shaping instrument. The EU has used regulatory standard-setting before to project influence beyond its borders, and there is every indication it intends to do the same here.
The United States, by contrast, has oscillated between executive orders, voluntary commitments, and agency-level guidance that carries no statutory weight. The current administration has signaled skepticism toward prescriptive rules, preferring what officials describe as an "innovation-first" posture. What that means in practice is that the federal government is largely asking the industry it is meant to oversee to describe its own risks and propose its own mitigations.
That is not governance. That is a negotiation in which one side controls all the relevant information.
## The Coordination Problem Nobody Wants to Solve
Beyond the U.S.-EU dynamic, a broader fragmentation is accelerating. Countries across the Global South are building their own frameworks — sometimes with UNESCO guidance, sometimes through bilateral arrangements with major AI exporters — that reflect genuinely different priorities around data sovereignty, labor displacement, and state capacity. Georgia's phased roadmap, for instance, reflects the practical reality that a country with limited regulatory infrastructure cannot simply import a European compliance architecture wholesale.
This is not inherently wrong. There are legitimate reasons why AI policy should reflect local contexts. But fragmentation creates arbitrage opportunities. When rules differ significantly across jurisdictions, well-resourced actors can route their highest-risk deployments through the most permissive regimes, while lobbying in stricter ones to water down requirements before they take effect. The cost of that arbitrage is borne by the people those rules were meant to protect.
## Transparency as a Minimum Standard
What's largely missing from the current governance conversation is a commitment to baseline transparency that doesn't depend on jurisdictional alignment. Before governments agree on what AI systems should be allowed to do, they could agree on what information about those systems must be disclosed — to regulators, to affected communities, to the public. Audit trails. Incident reporting. Model documentation that is actually legible to non-specialists.
This is not a radical proposal. Financial regulators have required disclosure as a precondition for oversight for decades. The argument that AI is too complex or too fast-moving to permit meaningful transparency is largely made by those who benefit from opacity.
## What Comes Next
The fracturing of global AI governance is unlikely to resolve itself through goodwill. It will resolve — if it resolves — through some combination of a high-profile failure that forces political attention, a dominant regulatory model that others adopt or react against, or a multinational agreement that sacrifices ambition for consensus.
None of those outcomes is guaranteed to produce accountability. All of them are more likely to produce it if civil society, journalists, and affected communities are paying close attention before the crisis arrives, not after.
The window for shaping these frameworks is narrowing. Deployment is not waiting for governance to catch up.