The most consequential vote on artificial intelligence in 2024 will not be cast. The United States Senate and House, compressed by an election calendar and convulsed by a disagreement so fundamental it cannot even be framed as a debate, have effectively handed the global rulemaking pen to Brussels. I have watched this paralysis before โ in crypto, in the autumn of 2022, when regulators who had spent two years threatening action suddenly discovered they had nothing but press releases. The whispers are identical: "no time," "no consensus," "let the market sort itself out." Over the final three months of the EU AI Act's passage through parliamentary procedure, I tracked something telling: American technology companies quietly began assigning attorneys to draft compliance memos for a law their own government refused to write. The signal is not in votes cast. It is in documents prepared for a vote that will never happen. When the flow stops, we see what truly holds.
The structural picture deserves precision, because the stakes are not merely legislative. United States AI governance rests today on Executive Order 14110, issued in October 2023 โ a document any subsequent president can rescind with a stroke. It imposes reporting requirements on large-scale compute clusters and safety tests, but it lacks the binding character of statutory law. No congressional mandate underpins it; no appropriation constrains its enforcement. Across the Atlantic, the EU AI Act has cleared final parliamentary hurdles, establishing risk-tiered obligations โ data governance, documentation, human oversight, systemic risk monitoring โ for any provider serving European users, regardless of where that provider is incorporated.
The GDPR precedent is unambiguous: European rules do not merely regulate Europe. They become the reference standard for global compliance, because markets are expensive to relinquish and legal teams prefer one standard to twenty. Meanwhile, state legislatures in California and New York have begun advancing their own AI bills, threatening a patchwork of obligations across every major commercial hub. So the map is drawn: one jurisdiction with a law, one with an executive decree, fifty potential rulebooks, and a court system about to become the real author of policy.
This fragmentation is not an accident of schedule; it is the product of a deeper disagreement about what AI is. In my current work on cross-border payments, I analyze how capital selects routes through jurisdictions with clear settlement rules and avoids corridors with contested authority. The same logic governs trust: institutional capital does not flow to gray zones. It seeks an address for risk. Brussels has provided one. Washington has not.
This is a map I recognize from the crypto asset class. The United States has spent three years without a comprehensive digital asset market structure bill, while the European Union operationalized Markets in Crypto-Assets Regulation and moved forward. The SEC and CFTC continue to dispute jurisdiction over tokens that neither agency can cleanly classify; Congress has produced a cascade of discussion drafts โ FIT21 and its successors โ but not a statute. AI is not the first technology where Washington has mistaken disagreement for deliberation. It is merely the most consequential. The cost of that mistake compounds differently, however, because AI's rulebook, once written, will govern not just markets but the production of knowledge itself.
The first error is to believe that an absence of legislation means an absence of law. Regulators do not need statutes when they have lawsuits. The litigation environment surrounding AI is already shaping behavior more effectively than any bill could. The New York Times' copyright action against OpenAI โ and the wave of authorship and data-provenance claims trailing behind it โ is functioning as de facto rulemaking, determining what training data may be used, what constitutes fair transformation, and what documentation will be required when disputes arrive. Courts are slow, expensive, and unpredictable instruments of policy, but they are instruments nonetheless. Fragility is the price of unsecured innovation โ and in this vacuum, fragility is distributed unequally.
There is a second channel pretending to be a governance structure, and it is arguably more dangerous than open silence. In 2023, leading frontier laboratories gathered at the White House to pledge voluntary safety commitments โ red-teaming, watermarking, disclosure practices. These pledges made for excellent press releases and almost no enforceable law. From my auditing experience, I have learned to distrust any system whose safety depends on the goodwill of the party whose growth depends on shipping faster than its competitors. Voluntary commitments are the corporate equivalent of an uncollateralized loan: persuasive at origination, catastrophic at default. They do not bind successor management teams, do not survive a change in competitive pressure, and offer no remedy to a user harmed by a model failure.
I learned this lesson auditing lending protocols during DeFi Summer in 2020. For three weeks, I traced undercollateralized risk positions and watched yield incentives decouple from any real revenue base. My report, "The Sustainability Illusion," concluded that the absence of rules does not create freedom; it creates exposure. The protocols that survived the 2022 collapse were not the most innovative; they were the most conservatively structured โ the ones that had treated the possibility of litigation and liquidation as design constraints rather than afterthoughts. The same filtering mechanism is now operating on AI companies. Those with legal war chests will weather the copyright storm. Those without them will be acquired at distressed valuations or simply vanish. But the litigation channel has a structural ceiling. Courts resolve cases; they do not build systems. And what this moment demands is not resolution but architecture โ a way to make claims about AI systems verifiable.
This brings me to the valuation question, where the macro picture sharpens. From my seat as a cross-border payment researcher, and from the work I did in 2024 analyzing Bitcoin ETF flows for a European institution, I have come to see regulatory clarity as a form of liquidity. Capital does not ask whether a technology is good; it asks whether the risk is priced. In the first three months after spot Bitcoin ETF approvals, my data showed $12 billion in net inflows correlating with reduced volatility in traditional markets. The mechanism was not enthusiasm. It was legibility. Once a risk can be named, measured, and hedged, capital floods in. The absence of an AI law does not create an absence of risk; it creates unpriceable risk โ which is worse for capital formation because it demands an undefined premium. The compliance-cost discount is already visible in the private markets I track: European investors are quietly requiring AI startups to demonstrate EU AI Act readiness before term sheets are signed, an unofficial tariff imposed by uncertainty.
The market is already bifurcating along these lines. Short-horizon venture investors read the legislative stall as a deregulation dividend: faster product launches, lower compliance overhead, unfettered data acquisition. Their models project increased free cash flow and mark valuations up. Long-horizon institutions and government-contract vendors read the same stall as an open-ended liability: litigation exposure, state-level fragmentation, the possibility that a future administration overcorrects with a punitive law drafted in haste. Their risk models demand a discount. Both readings are rational. Neither can be arbitraged away, because the divergence is the market's honest expression of a genuinely unknowable variable. And here the crypto economy enters the frame โ not as an escape, but as a foundation.
The EU AI Act's requirements read, to a blockchain researcher, like a specification document for verifiable infrastructure. Consider what it demands: risk management frameworks, documentation of training data, audit trails of model behavior, transparency about sources, human oversight mechanisms for high-risk systems. Every one of these obligations is, at root, a requirement to prove something โ and proving things is what cryptography does. This is the convergence I have spent 2026 modeling with a small team of ethics and engineering researchers: the emergence of verifiable compute markets, where AI inference arrives with cryptographic attestation, where training provenance is anchored to immutable ledgers, and where auditing does not require trusting a corporate press office but the structure of public evidence.
When I built the incentive models for these markets, I projected a $500 million opportunity for verifiable data sources by 2028. That was before the regulatory stakes were clear. The convergence of blockchain's proof machinery with AI's accountability crisis is not a niche interest; it is the only answer to the question neither Washington nor Brussels has fully articulated: how do we trust a system we cannot inspect? The answer is that inspection must become infrastructure. The audit trail must be embedded in the technology itself.
But I must be honest about the industry I work within, because the same structural disease that afflicts Washington's AI governance afflicts our own ecosystem. We have dozens of Layer2 networks now, each claiming to scale Ethereum, each hoarding a slice of an already-thin user base. This is not scaling; it is the slicing of already-scarce liquidity into fragments, a process that benefits infrastructure developers and their venture backers while leaving users to hop between bridges and pray for the safety of cross-chain protocols. The fragmentation of the crypto economy is the mirror image of the fragmentation of American AI governance โ the same confusion of proliferation for progress, the same substitution of marketing narratives for structural integrity.
The regulatory vacuum does not rescue crypto from irrelevance in the AI story; it makes crypto's tendency toward fragmentation catastrophic. In a world demanding verifiable proof, a fragmented, duplicated, siloed blockchain ecosystem cannot deliver. What is required is not another chain; it is a common standard for attestation โ a layer where proofs can be verified regardless of where they were generated. Without that standard, the verifiable data market I projected will go to centralized vendors who ship compliance products faster than the decentralized world can coordinate standards.
The comfortable narrative in crypto circles is that Washington's failure is our opportunity โ that decentralized governance will rise where centralized regulators fear to tread. This is fiction, and it is expensive fiction. Decentralization does not eliminate the need for enforcement; it fragments accountability into a form no court recognizes and no enterprise can rely on. The EU AI Act will not be satisfied because a model's training log is visible on a public ledger. Standards are not technologies; they are agreements backed by coercion. The likely trajectory for American AI firms is not resistance but absorption: they will adopt Brussels' rules as internal best practice, because European markets are too large to abandon. They will discover that a rule written in Brussels is cheaper to implement than six rules written in six state capitals, each with its own definitions, enforcement apparatus, and appetite for precedent.
Neither should we ignore the third chair at the table. Beijing has already implemented its own tiered regulatory approach to generative AI, with algorithmic filing requirements and content accountability rules that predate any American statute. China's effort to shape global AI governance through multilateral platforms may be less visible to Western audiences, but it is visible to every non-aligned economy weighing which standard to adopt. The American narrative assumes a binary contest between Washington and Brussels. That is Western-centrism of the most expensive kind. The real competition is three-cornered, and the prize is not the fastest model but the default rulebook.
What does that mean for crypto? It means the decoupling thesis fails. AI tokens will not decouple from AI regulation; the two will converge under a single requirement โ verifiable trust. Crypto's real role is not to replace regulation but to become the compliance infrastructure regulation demands. DeFi's glass house shatters under its own weight when it pretends otherwise. A compliance infrastructure that six different chains are building six different ways is not infrastructure; it is a regulatory incident waiting to happen.
The United States Congress will not act this year. Brussels has acted, and its shadow now reaches every American boardroom. In the quiet aftermath, only the resilient remain โ and resilience in this cycle belongs to builders who treat verifiable accountability as a design constraint before any law demands it. Not because any single law will last, but because the current of trust flows toward structures that can prove themselves. Beyond the illusion, the current never truly stops. The architects who understand this will not wait for the vote; they will build the proof. When the next cycle turns โ and it always does โ they will be the infrastructure others run on. That is the only position worth occupying in a silence that speaks louder than any hearing transcript.