The American Arbitration Association announced Wednesday a coalition of 15 major technology and crypto companies to launch the Legal Context Protocol, an open standard that attaches enforceable legal terms to transactions conducted by autonomous AI agents.
The AAA, which has handled commercial dispute resolution since 1926, partnered with Integra Ledger to develop LCP as agentic AI systems move from research labs into production commerce. Founding contributors span enterprise tech and crypto: Google, IBM, Circle, Wayfair, the Stellar Development Foundation, Ava Labs, Cardano, Hedera, Crossmint, the Aptos Foundation, Sei Labs, and Mysten Labs, the original contributor to Sui.
The timing is deliberate. Gartner projects agentic payments will reach $15 trillion in spending by 2028, a figure that makes the current legal vacuum around agent-to-agent transactions look increasingly untenable.
Click-Through Agreements Don’t Work When Nobody Clicks
Bridget McCormack, president and CEO of the AAA, framed the problem bluntly during a May podcast discussing the protocol. “The legal infrastructure that has supported e-commerce over the last 20 years… like click-throughs and terms of service, none of that translates… when agents are negotiating with other agents,” she said.
The observation cuts to a genuine gap. When a human buys something online, they encounter a terms-of-service checkbox. Courts have spent two decades building doctrine around whether that click constitutes informed consent, what happens when terms conflict, and which jurisdiction governs disputes. The framework is imperfect but functional.
None of it applies when your calendar AI negotiates a meeting-room rental with a building’s facility-management agent, or when your purchasing bot locks in a supplier contract with another company’s procurement system. There is no click. There may not even be a single moment of human review before value changes hands.
“There had to be some understanding about how legal context attaches to agentic transactions,” McCormack explained.
LCP attempts to fill that gap by making three things machine-readable and verifiable: the legal terms governing a transaction, the consent mechanism that authorized it, and the dispute-resolution pathway if something goes wrong. The protocol doesn’t dictate what those terms must be. It provides a standard format so agents can discover, parse, and agree to them before executing.
David Fisher, CEO of Integra Ledger and a co-founding partner on the project, emphasized the sequencing problem. “Payment infrastructure is actively being built for AI agents. The legal layer, what was agreed, under what terms, and how disputes will be resolved, is not.”
The protocol complements existing work on agent payment rails, including the x402 standard and Machine Payments Protocol, by answering questions those systems don’t address: under what terms did this transaction occur, what law governs it, and what recourse exists if the outcome disappoints.
Google and PayPal’s Agentic Payments Protocol launched earlier this year with 120 partners focused on the mechanics of moving money. LCP sits on top of payment rails rather than replacing them, handling the contractual wrapper rather than the funds flow.
Crypto Projects Bet Early on Agentic Infrastructure
The founding contributor list skews heavily toward blockchain networks, which might seem surprising for a protocol that explicitly doesn’t require a blockchain to function. The AAA confirmed LCP is chain-agnostic by design.
The crypto presence makes more sense when you consider what these networks are actually building. Hedera has positioned its hashgraph as infrastructure for enterprise applications that need auditability without full decentralization. Mance Harmon, Hedera’s co-founder, articulated the stake plainly: “As AI agents start making decisions and transacting on our behalf, we need to know there’s a clear answer to what happens if something goes wrong.”
Stellar, founded to facilitate cross-border payments, sees agentic commerce as a natural extension of its remittance infrastructure. Cardano and Aptos have both invested in formal verification and smart contract tooling that could integrate with machine-readable legal terms. Ava Labs, the company behind Avalanche, has pushed subnet customization that could accommodate jurisdiction-specific compliance requirements.
Circle, the issuer of USDC, brings stablecoin settlement infrastructure to the table. If agents transact primarily in stablecoins (a reasonable assumption given the need for predictable value), Circle’s presence signals the protocol might integrate with programmable-dollar mechanics.
Integra Ledger’s role provides the identity piece. The company builds middleware that gives AI agents verifiable identity, a prerequisite for any legal framework. You can’t hold an agent to a contract if you can’t establish which agent made the commitment and on whose behalf.
The enterprise names (Google, IBM, Wayfair) suggest the protocol is targeting commercial adoption rather than consumer-facing use cases first. That makes sense. Businesses already spend heavily on contract management and dispute resolution. An autonomous purchasing agent that can negotiate terms within company-defined parameters and bind the company to enforceable agreements would deliver immediate efficiency gains.
The Trillion-Dollar Predictions Come With Asterisks
Market sizing for agentic AI varies enough to suggest nobody really knows yet. The Gartner projection of $15 trillion in agentic payment spending by 2028 implies agents will be handling roughly 15% of global GDP in payment volume within two years. That number would require adoption curves steeper than anything seen in enterprise software.
Digital Applied offered a narrower estimate in March: the agentic AI market growing from $7.6 billion today to $236 billion by 2034. That’s a 30x increase over a decade, aggressive but within the range of previous platform shifts. McKinsey’s global projections push to $5 trillion by 2030, splitting the difference.
Goldman Sachs researchers in May predicted agentic AI would drive a “24-fold increase in token consumption by 2030” as consumers and enterprises adopt the technology. (Token here refers to AI model tokens, the computational units that measure how much reasoning an agent performs, not cryptocurrency tokens.)
The dispersion tells you something: the base rates for agentic commerce don’t exist yet, so forecasters are extrapolating from different assumptions about adoption speed, use-case breadth, and average transaction size. The one point of agreement is that the trajectory points sharply upward.
For legal infrastructure, the absolute number matters less than the growth rate. If agent-to-agent transactions are doubling annually, the gap between “no legal framework” and “something went wrong” shrinks fast. The AAA is betting that enterprises will want dispute resolution in place before the first high-profile agentic failure lands in court.
The protocol arrives as regulators globally are still working through how to handle AI accountability in general. The European Union’s AI Act imposes requirements on high-risk AI systems but doesn’t specifically address inter-agent contracts. US agencies have issued guidance on AI liability but nothing specifically targeting autonomous commercial transactions. By establishing an industry-led framework first, LCP could shape how regulators eventually approach the space.
A parallel exists in early e-commerce. Industry groups developed standards for digital signatures, transaction records, and online dispute resolution in the 1990s. Those standards influenced legislation like the E-SIGN Act and the Uniform Electronic Transactions Act, which codified electronic contract enforceability. LCP might be positioning for a similar trajectory: establish adoption, then seek regulatory recognition.
Second-Order Effects Worth Watching
The protocol creates potential leverage for whoever controls its adoption. The AAA has a natural interest: if LCP becomes standard, the organization becomes the default arbitrator for the fastest-growing category of commercial disputes. That’s a substantial revenue line.
For crypto networks, LCP integration could become a competitive differentiator. A chain that natively supports LCP-compliant transactions might attract enterprise deployments over chains that require middleware. The founding contributor list reads like a bet on which networks will capture the agentic-commerce vertical.
Google’s involvement creates doubt about centralization. The company already dominates consumer AI through its products and provides cloud infrastructure to many of the agents that will transact. If Google shapes the legal framework those agents use, it gains another layer of platform control. The open-standard design theoretically prevents lock-in, but standards often benefit the organizations with resources to define them.
The insurance industry is watching too. Agentic transactions create novel liability questions: if an AI agent makes a catastrophically bad purchase, who pays? The principal who authorized the agent? The agent’s developer? The infrastructure provider? LCP doesn’t answer these questions directly, but verifiable transaction records and clear dispute pathways make actuarial modeling possible. Expect insurers to engage with the protocol as agentic adoption scales.
One risk the protocol doesn’t address: adversarial agents. If an AI system is designed to manipulate rather than cooperate, LCP’s consent mechanisms could become attack surfaces. An agent that can convincingly appear to agree to terms while actually operating under different constraints creates enforcement nightmares. The protocol assumes good faith in ways that may not survive contact with motivated adversaries.
Mance Harmon’s comment points toward the core uncertainty. When something goes wrong between AI agents, we need clarity on consequences. LCP attempts to provide that clarity by making legal terms machine-readable and arbitration accessible. Whether the framework holds up when tested by real disputes, large sums, and adversarial actors remains the open question.
The crypto industry’s early involvement suggests a belief that programmable money and programmable contracts belong together. If agents transact in stablecoins over smart-contract rails with LCP-compliant legal wrappers, the stack becomes genuinely different from traditional commerce, not just faster but operating under different assumptions about automation, verification, and enforcement.
That’s the bet embedded in today’s announcement: that the legal infrastructure for machine commerce needs to be built now, before the machines are fully autonomous, and that whoever builds it shapes how trillions in future transactions flow.
