XRP Ledger patches decade-old bug that could have minted trillions of XRP

XRP Ledger developers said Friday that a bug in the network's payment engine could have let an attacker create XRP out of nothing. The critical-severity flaw had been in the code since 2015, according to a disclosure report, which said there was "no evidence that this issue was exploited on any public network."

The bug affected how the engine totals a payment that fills many order book offers at once. Those sums had no overflow check, so a large enough total wrapped around to a small number.

An attacker could have set up a few hundred accounts, have each of them list a tiny fraction of a token for sale in exchange for an enormous amount of XRP, then close all of the sales with one payment, per the report. Each seller wallet would have gotten paid in full, while the buyer wallet would've been charged "only a few hundred drops plus the fee." The network's check against newly created XRP had a similar issue adding the balances, and thus would fail to prevent the minting. 

"The attack could not be triggered by accident," the report states. "It required hundreds of offers priced in a way no real trader would use, followed by a payment built specifically to consume them all at once."

One transaction, 18 trillion XRP

Cayden Liao, security co-founder at Veria Labs, wrote on X Saturday that the firm's AI agent found the bug and built a working exploit. By his count, just one exploit transaction on the XRP Ledger could have created about 18 trillion XRP, 184 times the token's total supply, which was capped at 100 billion XRP at launch and shrinks slightly as transaction fees are burned.

"We suspect one reason nobody caught this vulnerability is that it chains together two bugs that would be low severity on their own," Liao wrote. 

Liao said Ripple paid Veria the program's top $250,000 bounty. "To our knowledge, that's the largest ever paid out for a vulnerability discovered entirely by an AI agent," Liao added.

Skipping the amendment vote

Veria reported the bug on Sept. 22, and RippleX shipped a fix in xrpld version 3.4.1 three days later. More than 80% of validators on the default Unique Node List were running it that day, the report said.

Rule changes on the XRP Ledger normally go through an amendment process, with each proposed change requiring support from more than 80% of the network's trusted validators over a period of two weeks. However, this bug fix took effect as each server upgraded, which the report said was a first in more than 10 years of amendments. Because xrpld is open source, a normal release would have shown attackers where the bug was for weeks before the fix switched on, the report said.

The release also patched a separate bug in Batch, a feature that had not gone live when the flaw was found, which the report said could have stalled the network. Batch and its fix both activated Friday, and servers on older versions can no longer stay in sync.

For the Batch bug, a potential attacker "does not gain anything on-chain," the report said. "The payoff is a permanently malformed record and the chance to break off-chain indexers, explorers, and SDKs."

Closed-source criticism

The source code for xrpld's 3.4.1 update was published about two weeks after the binaries. Cyber Capital founder Justin Bons wrote on X Thursday that "XRP has been running closed-source code for the last two weeks."

"AI has changed things," RippleX engineer Mayukha Vadari replied on Saturday. "You can't sneak a critical bug patch into a routine public release process, it'll get spotted and reverse-engineered immediately."

RippleX Head of Engineering J. Ayo Akinyele said in a separate X post on Saturday that the team would expand AI-assisted bug hunting and formal verification. "AI is fundamentally changing how quickly vulnerabilities can be discovered," he wrote.