Ripple Deploys AI to Bolster XRP Ledger Security for Institutional Use

AI agent payments are an emerging form of infrastructure that allows software agents to initiate, approve and settle transactions autonomously. Ripple is bringing the XRP Ledger, or XRPL, into this market, aiming to support cross-border payments with XRP and the RLUSD stablecoin while competing with Visa, Mastercard, Coinbase and other players.

As of July 20, 2026, Ripple had launched the XRPL AI Starter Kit, enabling developers to build AI agents that can automatically approve and settle payments. Agent payments currently remain dominated by USDC. Ripple has not disclosed the investment associated with the kit, transaction fees or the scale of any formal commercial deployment.

The XRP Ledger (XRPL) is a public blockchain supporting payments and tokenized assets. The proposed Batch amendment, which was awaiting a vote, was designed to allow multiple transactions to be executed in batches. A bypass of its signature verification could have allowed attackers to make payments from any account without obtaining its private key, putting assets across the network at risk and underscoring the value of AI-assisted code audits.

On February 19, 2026, Apex, an autonomous tool developed by Cantina AI, discovered the vulnerability. Ripple's engineering team verified it that evening, and validators on the Unique Node List (UNL) were immediately notified to vote against the amendment. Rippled 3.1.1 was released on February 23, disabling Batch and fixBatchInnerSigs. The flaw never reached mainnet and caused $0 in losses, while the replacement proposal, BatchV1_1, remains under review.

The XRP Ledger, or XRPL, is a public blockchain supporting XRP and on-chain transactions, while the XRP Ledger Foundation, or XRPLF, promotes its technology and ecosystem. The vulnerability affected signature-verification logic in the yet-to-be-activated Batch amendment. It could have allowed attackers to execute transactions on behalf of accounts and steal assets without their private keys. Cantina’s CEO estimated the direct exposure at nearly $80 billion.

Cantina’s Pranamya Keshkamat and its AI bug-hunting tool Apex discovered the flaw through static analysis and reported it on February 19, 2026. The amendment was still undergoing a validator vote and had not been activated on the mainnet, so no funds were lost. XRPLF advised validators to vote against it and released the emergency rippled 3.1.1 update on February 23, preventing the amendment from being activated.