Two major distributed ledger ecosystems have outlined frameworks for keeping human oversight and cryptographic accountability at the center of AI-driven payments. XRP Ledger documentation published on 4 October 2026 describes how developers can build approval checkpoints into autonomous payment agents, while Hedera Chief Policy Officer Nilmini Rubin called for cryptographic identity, verifiable audit trails, predictable fees and deterministic finality in a piece for OMFIF's Digital Monetary Institute Journal on 5 October 2026.
The XRPL guidance focuses on a practical workflow: an AI assistant preparing a payment must hand a proposed transaction to a wallet skill for signing only after a human reviews the full recipient address, amount, network and fee. The XRPL Payments skill lets an agent construct transactions for XRP and RLUSD transfers. Automatic signing can be enabled, but only with an explicit temporary scope that includes transaction type, network, expiry, approved destinations and amount caps. The documentation warns that a per-payment cap does not create a total budget, and it highlights prompt injection as a key threat: incoming transaction memos and invoice instructions are treated as untrusted input and cannot authorize payments on their own.
The XRPL materials also distinguish between signing configurations. A local environment-variable seed may be suitable for development or low-value accounts, while an external signer keeps keys outside the agent process. The Open Wallet Standard vault with a scoped, revocable agent token is presented as the more consequential production option because it triggers policy checks before signing, whereas sharing an owner vault passphrase would grant full access and undermine intended limits.
On Hedera, Rubin argued that traditional know-your-customer assumptions fail when AI agents can create payment accounts at scale and act for multiple principals across jurisdictions. She proposed cryptographic identity standards linking agents to the people or institutions behind them, with smart contracts enforcing spending limits and clearly defined operating boundaries. Rubin cited EQTY Lab's Verifiable Compute as an example of infrastructure that generates cryptographic certificates for AI operations using NVIDIA and Intel hardware, with attestations anchored through the Hedera Consensus Service. Hedera, Accenture, and EQTY Lab have also developed a public-sector architecture combining Verifiable Compute, Accenture's multi-agent gateway, NVIDIA DGX Cloud, and Hedera Consensus Service.
Rubin also highlighted the x402 protocol, which allows agents to pay automatically for data and services without account setup or approval. Hedera has integrated x402 into its agent kit, and supporters include Google, Visa, Amazon Web Services and Anthropic. She said payment options within this framework include stablecoins and tokenized deposits, with tokenized deposits giving regulated banks access to programmable money under existing supervision.
The combined publications do not introduce protocol-level human-approval requirements, but they provide developer and institutional guidance for building safeguards around autonomous payments. Their ultimate effectiveness depends on implementation and adoption, while the broader agentic payment landscape remains subject to technical, regulatory and adoption risks.
