An AI-powered code analysis tool created by Jack Dorsey’s Block uncovered a security flaw that could temporarily disrupt Bitcoin nodes by forcing them to process the same transaction ID 50,000 times.
Bitcoin Core, the software that powers Bitcoin nodes, publicly disclosed the issue on Friday as CVE-2026-117348. Developers also rolled out versions 30.5 and 31.3, which contain patches designed to prevent the vulnerability from being exploited.
The security notice credited an automated investigation carried out using Project Loupe, an open-source vulnerability scanner developed by Spiral, Block’s Bitcoin-focused development unit. The project is led by Elias Rohrer, CoinDesk reported.
Bitcoin nodes are responsible for validating transactions and distributing transaction and block information among network participants.
Exploitation of the vulnerability could have created operational problems for cryptocurrency businesses. Exchanges using affected nodes might have experienced delays in confirming deposits, while miners could have continued processing outdated blockchain information and wasted computing power.
Bitcoin Core’s development team classified the issue as a medium-severity vulnerability that could be discovered relatively easily. Its security policy, introduced on October 22, requires developers to release fixes across maintained software versions as quickly as possible when vulnerabilities are easy to identify. Public advisories are also issued alongside the patches.
How the AI-Discovered Bug Could Disrupt Bitcoin Nodes
Bitcoin nodes store orphan transactions, which are payments that reference coins from earlier transactions that have not yet reached the receiving node. To process these transactions, the software must identify and retrieve the missing parent transactions.
Since version 29.0, Bitcoin Core has generated a list of missing parent transactions whenever another node announces an orphan transaction. It checks its existing records before determining which information needs to be requested.
However, the system allows transaction announcements containing up to 50,000 entries and did not automatically eliminate duplicate IDs.
An attacker could exploit this behavior by submitting an orphan transaction with thousands of missing parent transactions and then broadcasting an announcement containing the same transaction ID 50,000 times.
This would force the receiving node to repeat computationally expensive checks for identical transaction information, potentially overwhelming its processing capacity.
Bitcoin Core’s advisory warned that a single malicious announcement could stop a node from handling other network messages for more than a dozen seconds. During this period, the node would also be unable to process incoming block announcements. Additional malicious messages could prolong the interruption.
The latest security patches eliminate this weakness by ensuring that each transaction announced in a batch is checked only once, regardless of how many duplicate entries it contains.
The discovery highlights the growing role of artificial intelligence in cryptocurrency security research. Although AI-powered scanners can help developers identify software vulnerabilities before attackers exploit them, these tools could also make security weaknesses easier for malicious actors to uncover.
Bitcoin Core’s rapid response reflects developers’ concerns that AI-assisted vulnerability discovery could reduce the effectiveness of keeping security flaws undisclosed while patches are being prepared.





