Political Tech
The Geneva moment has passed — and the governments tasked with governing artificial intelligence are running a distant second to the systems they're supposed to oversee.
NewsOnScale Staff
October 7, 2026
There is a version of the AI governance story that flatters everyone involved. In this version, world leaders are engaging seriously with the risks of advanced AI, bilateral summits are producing meaningful guardrails, and regulatory frameworks are catching up to the technology in something like real time.
That version is not the one playing out.
## The Summit That Raised More Questions Than It Answered
When representatives from the United States and China sat down to discuss artificial intelligence at their recent high-level summit, the conversation was historic in one narrow sense: two superpowers acknowledged, formally, that the technology their companies are racing to build might require some kind of mutual restraint. That acknowledgment matters.
But acknowledgment is not architecture. The substantive differences between Washington's preference for minimal regulatory friction and Beijing's insistence on state-level security controls over AI systems are not minor negotiating gaps — they reflect genuinely incompatible theories of what AI governance is *for*. The U.S. frame centers economic competitiveness and voluntary industry standards. The Chinese frame centers party oversight and what officials describe as "algorithmic sovereignty." These are not positions that split the difference easily.
What emerged from the summit, by most accounts, was a commitment to keep talking. That is not nothing. It is also not a governance framework.
## The Institutional Lag Is Structural, Not Incidental
Zoom out from the bilateral dynamic and the picture gets worse, not better. Across democracies — in Europe, in the Indo-Pacific, in the Americas — the pattern is consistent: legislative bodies are debating frameworks for AI systems that have already been superseded by newer deployments. Regulatory agencies tasked with AI oversight frequently lack the technical staff to audit the systems they're nominally overseeing. And the procurement pipelines pushing AI into courts, hospitals, schools, and welfare offices are often moving faster than the policies governing their use.
This is not primarily a funding problem, though funding is part of it. It is a structural mismatch between the timelines of democratic deliberation — which are long, iterative, and appropriately cautious — and the deployment timelines of AI development, which are compressed, compounding, and driven by competitive pressure that no single government controls.
The result is a kind of governance theater: frameworks get announced, working groups get convened, principles get published. Meanwhile, the systems that actually affect people's lives — the hiring algorithms, the content ranking engines, the automated decision tools embedded in public services — operate largely outside meaningful accountability structures.
## Who Fills the Vacuum
When governments lag, someone else sets the norms. In practice, that means the largest AI developers are writing the de facto rules of the road, with voluntary commitments serving as the primary check on deployment decisions that affect billions of people. It also means that the international governance conversation is being shaped disproportionately by actors — the U.S., China, and to some extent the EU — with the resources to participate, while smaller nations are largely rule-takers in a system they had little hand in designing.
This dynamic has a name in trade and financial governance: regulatory capture by scale. The entities large enough to engage in the rulemaking process are, by definition, the entities with the most to gain from shaping the rules in their favor.
## What Accountability Actually Requires
None of this means governance is impossible. It means the current approach — summits, voluntary frameworks, aspirational principles — is insufficient on its own. Meaningful AI governance requires governments to build internal technical capacity, not just convene external advisory panels. It requires procurement rules with teeth, not just guidance documents. And it requires transparency mechanisms that let the public actually see how automated systems are being used in their name.
The conversation that started in Geneva, and continued in recent U.S.-China talks, is a necessary beginning. But the institutions capable of turning those conversations into enforceable accountability haven't been built yet. Until they are, the gap between AI's reach and democracy's grip will keep growing.