Two thousand proposals and no framework

Rules aimed at named harms are easy to draft and easy to pass. That is why there are thousands of them, and why none of them adds up to a regime.

More than 2,000 AI proposals are in play and not one establishes a durable regulatory architecture. Meanwhile a judge denied xAI's bid to block Minnesota's synthetic-imagery ban — partly because the company waited until three days before the law took effect.

Specific harms are tractable. Frameworks are not.

Deepfakes, hiring discrimination, chatbot disclosure — each has an identifiable victim, a describable mechanism, and a constituency. Bills like that pass.

A framework requires agreeing what an AI system is, who is accountable when one causes harm, and which regulator has jurisdiction. None of those has an answer that survives a committee, which is why two thousand instruments have produced no architecture to hang them on.

Who the patchwork favours

Compliance cost that scales with jurisdiction count rather than risk level is a structural subsidy to whoever can afford fifty simultaneous compliance programmes. That is not the outcome any individual bill intended and it is the outcome all of them together produce.

The courts are not filling the gap

The Minnesota ruling turned on timing, not substance — a temporary restraining order needs imminent harm, and a company that waits until the eve of enforcement has argued against its own urgency. Nothing constitutional was decided.

That is the pattern. Courts are declining to move fast on AI, which means the legislative traffic matters more than the litigation, and the legislative traffic is two thousand pieces that do not fit together.

The credible counter-argument

There is a serious position that AI can be governed through existing sectoral law — that a discriminatory hiring decision is already illegal regardless of what made it, and no new acronym is required.

It is stronger than it sounds, and it may be why the gap persists by choice rather than failure. But it has a hole, and an open-weight model that refuses nothing and has no published safety framework is sitting in it. Sectoral law reaches deployers. It does not reach a weights file.

CNBC — Over 2,000 proposals aim to govern AI. Not a single one addresses a long-term regulatory framework → · TechCrunch — Judge denies xAI's request to block Minnesota ban on nudify apps → · IAPP — No new acronyms required: governing AI without 'AI law' →