Ripple's 10,000-Line Trim Shows Security Isn't Just About Audits

Ripple's got a simple thesis right now: less code, fewer attacks. The company wants to pull the plug on XChainBridge, a cross-chain protocol that never found its footing. Removing it and a related amendment would strip more than 10,000 idle lines from the XRP Ledger's core software. That's not just housekeeping. Every line of unused code is a place a bug can hide, a door an attacker can try.

Here's the thing. The bridge was originally designed to connect XRPL to its own EVM sidechain. But Ripple chose Axelar for that job back in June 2024 after weighing security, decentralization, and the sheer hassle of maintaining a custom bridge. The native option sat around collecting dust, kept alive for a validator vote that never turned into real demand. Developers got 12 to 15 months to prove projects needed it. They didn't. So now it's got a one-way ticket out of the codebase, subject to the amendment process, of course.

At the same time, Ripple is prepping the ledger's next big financial feature: Lending Protocol V1.1. On Aug. 27, Sherlock's Audit Engine started an AI-only security review of the system. That's a bold bet, given the stakes. The earlier lending code went through a $200,000 attackathon in late 2025 that pulled 455 submissions from 131 researchers. It produced 94 valid findings, 15 of them critical. Then Ripple's own AI red team flagged 20 lending-specific issues between March and May, including an inverted invariant that could've hid phantom collateral and an integer overflow that could've deadlocked a node.

So why trust AI alone now? Ripple's own researchers warn AI pipelines generate false positives and can misinterpret invariants. It's a test. Sherlock hasn't released findings or a completion date. The industry backdrop makes this urgent: $1.315 billion lost across 344 incidents in the first half of 2026, per CertiK. Code vulnerabilities appeared in 204 of those cases. Hackers are circling old contracts too, meaning yesterday's mistakes compound into today's breaches.

Cutting dead code is the cheapest security fix in existence. The AI review is the expensive, uncertain one. Ripple is doing both, which tells you where confidence actually sits. Not in any single tool, but in layers: audits, competitions, fuzzing, red teams, and human judgment. Payments, not speculation. That's the point. If V1.1 hits mainnet with a blind spot, we'll all know. Fast.