AI Audit at Block Uncovers Bitcoin Core Bug That Could Stall Nodes via Repeated Transaction ID

An AI audit at Jack Dorsey’s Block uncovered a Bitcoin Core flaw that could freeze node operations for seconds by repeating a single transaction ID 50,000 times, forcing an emergency patch on Friday.

The vulnerability, tracked as CVE-2026-117348, was found by Loupe, an open-source scanner developed at Block’s Spiral unit that uses AI agents to examine bitcoin code.

Bitcoin Core released versions 30.5 and 31.3 to address the issue, which developers rated as medium severity and “easily discoverable.” The advisory credits an automated AI scan conducted through Elias Rohrer’s Project Loupe at Spiral.

Under a security policy adopted Oct. 22, easily discoverable flaws receive fixes across maintained versions as soon as possible, with a public advisory alongside the releases. The retired 29.x branch will not receive a fix, according to The Defiant.

Bitcoin nodes hold orphan transactions — payments that spend coins from an earlier transaction the node has not yet seen. Since version 29.0, when another node announces such an orphan, Bitcoin Core builds a list of its missing parent transactions and checks its records before deciding what to request. Nodes announce transactions in batches of up to 50,000 entries, and the software did not skip repeated entries.

An attacker could first send an orphan transaction with thousands of missing parents, then repeat its ID 50,000 times in one announcement. The receiving node would perform the expensive checks over and over. “Each such message stalls the node for over a dozen seconds,” the advisory said. During that time, it processes no messages from other nodes, including new blocks. Sending further messages could prolong the disruption.

An exchange relying on an affected node could be delayed in crediting deposits, while a miner could waste computing power working on an outdated view of the blockchain. The fix makes the node process each transaction announced in a batch only once.

The policy reflects developers’ concern that AI tools make some vulnerabilities cheap for attackers to rediscover, reducing the protection offered by keeping fixes quiet. Whether the Loupe finding marks a broader shift in how quickly such flaws will be surfaced remains an open question.

Photo of author

Newsroom

Newsroom at Archyde.

Brighton Second After Nine Games; Beat City With £25m Midfield vs £327m Trio

Leak: Samsung Developing ‘Moment Effects’ Dynamic Island Feature, Could Arrive With Galaxy S27