Korean policymakers are moving from observation to rulemaking on autonomous machine commerce. The National Assembly this week pointed out that although the era of AI agents signing contracts and even making payments on behalf of people has effectively arrived, the legal framework for who is responsible when a machine commits funds remains undefined. The Korea Blockchain Industry Promotion Association has responded by convening an AI Trust Infrastructure Policy Forum on September 11, aiming to anchor the discussion in blockchain-based identity and accountability rather than leaving it to the card networks.
What the Assembly Actually Said
The parliamentary critique, reported by Maeil Business, is straightforward: agents are already executing transactions, but contract law presumes a human principal with capacity, intent and liability. When an agent negotiates and pays autonomously, existing statutes do not clearly allocate responsibility between the user, the agent operator and the payment provider. This is not a hypothetical in Korea’s framing — agents purchasing services and settling in digital assets is treated as a present-tense phenomenon that has outrun the law. The Assembly’s intervention signals that Korea intends to treat agent payments as a regulatory category of its own rather than folding it into generic AI safety or fintech rules.
The September 11 Forum
The Korea Blockchain Industry Promotion Association’s policy forum is the industry-side counterpart. Its premise is that blockchain infrastructure — verifiable identity, auditable transaction trails, programmable spending limits — is the natural trust layer for agent commerce. That framing matters because it positions public ledgers and stablecoin-style settlement as the accountability substrate, not merely the payment rail. If an agent’s authority, budget and transaction history live on-chain, disputes about unauthorized or erroneous payments have a shared evidence layer. The forum’s outcome is worth watching: Korean industry associations have historically fed directly into legislation, and a concrete proposal on agent identity would be one of the first from a major jurisdiction.
Why This Mirrors the Card Network Effort
We covered earlier this week that Visa, Mastercard and Ant International are building common standards for identifying and monitoring paying agents, focused on fraud and interoperability. The Korean discussion asks the question upstream of that work: even if you can identify an agent, who bears legal liability when it transacts? Card-network standards solve authentication; they do not by themselves answer contract validity, consumer protection or dispute resolution when no human approved the specific transaction. Korea’s approach — trust infrastructure plus legislative clarification — could end up more durable than network-level rules, which are enforceable only within closed ecosystems.
Stakes for Stablecoin Rails
For agent-native payment protocols like x402 and Coinbase Agent Payments, regulatory clarity is double-edged. On one hand, a statutory framework that recognizes machine-initiated payments removes the legal ambiguity that keeps enterprises cautious. On the other, if the rules mandate liability structures or licensing requirements designed around card rails, stablecoin-based agent payments could inherit compliance costs they were built to avoid. The detail to watch is whether the forum treats on-chain identity and spending limits as sufficient accountability, or demands a human-approvable fallback for every transaction — the latter would blunt much of the autonomy that makes agent commerce economically distinct.