AAA Warns Existing Law Cannot Handle Agent Commerce—and Courts Will Impose Their Own Rules

Agentic commerce—autonomous AI systems negotiating and executing transactions on behalf of users—is in direct conflict with a rigid, centuries-old legal framework. While developers race to build the payment rails that will facilitate these machine-to-machine exchanges, the American Arbitration Association (AAA) has issued a stark warning: existing contract and agency law is fundamentally ill-equipped to handle the reality of autonomous agents.

In a 2026 paper titled Agentic Commerce Needs Legal Infrastructure—and the Courts Are Coming, the AAA, the world’s largest private provider of alternative dispute resolution services, argues that the legal architecture for this new economy is missing. Bridget McCormack, President and CEO of the AAA-ICDR and former Chief Justice of the Michigan Supreme Court, joined David Hoffman, a professor at the University of Pennsylvania Carey Law School, on a July 2026 episode of the Ballard Spahr Consumer Finance Monitor to detail why market forces alone will not bridge this gap.

The core tension lies in the application of traditional legal doctrines to non-human actors. Concepts like clickwrap agreements, disclosure requirements, and informed consent were designed for human users. When an AI agent initiates a transaction, these mechanisms break down. The AAA paper highlights that existing agency law, including the principle of ratification, could inadvertently bind companies to contracts negotiated by their AI agents, even when those agents act outside of intended parameters.

The risk is not merely theoretical. The paper warns of mass-scale class-action litigation if AI agent errors are replicated across millions of autonomous transactions. If a single algorithmic flaw leads to unauthorized or erroneous purchases at scale, the resulting legal liability could be catastrophic. Proving what an AI agent actually agreed to—and maintaining a reliable, machine-readable record of that contract formation—remains a significant hurdle that current systems are not prepared to clear.

The current development of payment protocols prioritizes technical speed over legal alignment, creating a misalignment of incentives for stakeholders. The industry is currently engaged in a race between three primary architectures: the crypto-native x402 stablecoin settlement, Google’s UCP card-rails checkout, and Mastercard’s AP4M multi-rail system. While these protocols focus on the technical efficiency of moving value, they largely ignore the load-bearing nature of the legal gap. In traditional card networks, chargeback mechanisms provide a safety net for consumers. However, as commerce shifts toward irreversible payment systems like stablecoins, that safety net disappears. Without a robust legal infrastructure, a dispute over an agent-initiated transaction may have no path to resolution.

Paymentology CTO Tim Joslyn noted in July 2026 that 99% of existing issuer processing systems are already capable of handling agentic payments. This suggests that the competitive battleground is not the payment rails themselves, but the trust infrastructure—identity, authorization, and fraud prevention. Yet, consumer trust in these tools remains low. A Product.ai survey from April 2026 found that only 14% of consumers trust AI recommendations without verification, and AI tools ranked 6th out of 7 information sources in terms of consumer trust.

Regulatory bodies have yet to provide clarity. The rulemaking deadline for the GENIUS Act passed on July 18, 2026, with no final rules issued by any of the six agencies involved. Crucially, the Act does not specifically regulate AI agent-initiated transactions, leaving a vacuum where industry standards are expected to fill the void.

The market is already beginning to bifurcate in response to this uncertainty. A clear fork has emerged in how disputes might be handled: the crypto-native Internet Court, which utilizes 1,001 AI validators to resolve cases for under $1.50, versus the AAA’s own approach, which seeks to adapt the traditional legal system through machine-readable terms. The AAA’s institutional preparation is evident; the organization has already deployed an AI Arbitrator product, signaling that it expects the volume of AI-related disputes to rise significantly.

The industry’s focus on technical interoperability is understandable, but it introduces significant operational and financial risk. If the legal liability layer remains unresolved, the efficiency gains promised by agentic commerce may be eclipsed by the costs of litigation and the unpredictability of court rulings. As the AAA paper suggests, if the industry fails to build this infrastructure proactively, the courts will eventually impose their own, likely producing fragmented and costly outcomes. The race to build the next generation of payment protocols is well underway, but the most critical infrastructure—the rules of the game—remains largely unwritten.