One year after the passage of the Guiding and Establishing National Innovation for U.S. Stablecoins Act (Public Law 119-27), the federal government has failed to finalize a single rule. Despite a year of proposed rulemakings from the OCC, Treasury, and FDIC, the administrative momentum has stalled. As the deadline passes, the industry is left with a stack of NPRMs and a growing, unaddressed reality: the law was designed for a world of human-led finance that no longer exists.
The GENIUS Act was intended to provide the bedrock for stablecoin issuance. Its scope is strictly limited to the mechanics of the issuer: reserve requirements, redemption protocols, capital floors, and BSA/AML obligations. It is a framework for the pipes, not the traffic. While regulators spent the last year debating whether a new issuer needs a $5 million capital floor or how state-level oversight should interact with federal mandates, the actual usage of these stablecoins has shifted toward autonomous, machine-to-machine commerce.
The legislative design is entirely silent on AI agent-initiated transactions, programmatic authorization, and the liability that arises when an autonomous agent executes a payment without a human principal. This structural gap is far more significant than a mere delay in rulemaking. As David Krause noted in the University of Miami Law Review, there are no dedicated provisions governing these smart-contract-driven transfers. We are building a high-speed settlement layer for AI agents on top of a regulatory foundation that assumes every transaction is authorized by a human with a wet-ink signature or a digital equivalent.
The United States is currently operating in a vacuum, lacking even the preliminary guidance seen elsewhere. In the UK, the Competition and Markets Authority (CMA) has already established that the use of AI does not absolve a business of its consumer law obligations, effectively pinning liability for agent actions on the deployer. The UK’s Financial Conduct Authority (FCA) has gone further, explicitly flagging that existing payment regulations—specifically those requiring payer consent per transaction—are fundamentally incompatible with autonomous agent payments. While the UK is far from a finished framework, they have at least identified the friction points. In the U.S., Regulation E remains the standard, a relic written for human-authorized transactions that offers no clarity for the machine-led economy.
The industry is not waiting for the regulators to catch up. With stablecoin transaction volumes hitting approximately $33 trillion in 2025, these assets have become the default settlement layer for AI agents. Major players are already deploying infrastructure into this liability void. According to Visa company data, its Intelligent Commerce platform is seeing $7 billion in annualized stablecoin settlement, growing 50% quarter-over-quarter. Mastercard, Stripe, and the 40-member x402 Foundation are pushing ahead with machine payment protocols. These firms are building the rails for an autonomous economy while the federal government remains stuck in a cycle of proposed rules for traditional issuer mechanics.
No dedicated provisions govern AI-agent-initiated stablecoin payments, smart-contract-driven autonomous transfers, or liability allocation when an AI agent executes transactions without a human principal.
David Krause, 80 U. Miami L. Rev. 337 (2026)
By focusing exclusively on the issuer, the Act ignores the reality of how these assets are being used. When an AI agent executes a transaction that falls outside of its user’s intent, or when a micropayment settlement fails due to a logic error in an autonomous contract, there is no federal guidance to determine who bears the loss. The law regulates the stablecoin as a product, but it ignores the stablecoin as a programmable utility.
For builders and investors, the passing of the July 18 deadline is a signal to stop waiting for federal clarity on agent-based commerce. The GENIUS Act was never intended to solve the problem of autonomous liability, and the current regulatory silence suggests that the U.S. is content to let the courts or future, reactive legislation handle the fallout. Until then, the industry is operating in a high-stakes environment where the technology is moving at the speed of code, while the liability framework remains anchored to the human-centric assumptions of the past.