Bitcoin Core shipped patched releases 30.5 and 31.3 on Nov. 20, closing a medium-severity denial-of-service hole that let a single peer knock a node offline for stretches of more than a dozen seconds with a single crafted message. The vulnerability, tracked as CVE-2026-117348, was not turned up by a human code reviewer but by Project Loupe, an automated large-language-model scanner run by Elias Rohrer at Spiral, the open-source development arm funded by Jack Dorsey's Block.
The disclosure followed a patch-first sequence: fixed binaries were published before the technical details went public, narrowing the window in which an unpatched node could be attacked by someone reading the advisory.
What the flaw actually did
A Bitcoin Core node is the software that independently checks every transaction and block, relaying both across the network. When it stalls, it stops handling incoming data, so it cannot pass along transactions, confirm blocks or serve its peers until it recovers. That is a disruption rather than a compromise, but it can be retriggered at will by any peer willing to send the right message.
The weakness sat in how Bitcoin Core 29.0 and later handled announcements of orphan transactions, payments that spend coins from a parent the node has not yet seen. The node queues those missing parents for download. According to the official advisory, the orphan-announcement processing added in 29.0 had no deduplication logic.
A protocol inventory message can carry as many as 50,000 entries. By packing one transaction ID into all 50,000 slots, an attacker made the node treat every duplicate as a new request and repeat the same expensive lookups. The advisory states that each such message stalls the node for over a dozen seconds, and that messages sent in quick succession compound the effect, cutting the node off from legitimate blocks entirely.
"Each such message stalls the node for over a dozen seconds," the advisory said.
What the bug did not do matters as much as what it did. It did not permit theft of funds, double-spending or any change to the blockchain. Bitcoin's cryptographic integrity was never in question; the problem lived in the node software, not in the protocol.
Timeline of discovery and fix
Sanders validated the finding, opened the pull request and had the deduplication change merged two days after the independent confirmation arrived. The patch collapses repeated transaction IDs in a single announcement so each is processed once instead of triggering a separate lookup.
Every node running versions 29.0 through 31.2 was exposed. The 29.x branch has reached end-of-life and will not be patched, leaving operators on that line with an upgrade to a supported release as their only remedy.
A policy built for cheap discovery
The disclosure timeline reflects a rule change Bitcoin Core made on Oct. 22, when it began treating bugs rated "easily discoverable" as requiring accelerated public disclosure. The rationale is that if a language model can surface a flaw quickly and at low cost, an adversary running the same class of tool can too, so a traditional embargo buys little protection.
CVE-2026-117348 was classified "Medium / Easily Discoverable" under that framework, which is why the advisory appeared alongside the releases rather than months afterward. It is not the first time automated scanning has produced a Bitcoin Core fix: in October the project disclosed a separate set of bugs found the same way, including issues that could crash nodes on Windows and macOS.
What operators should do
Nodes on the 30.x line should move to 30.5; those on 31.x should move to 31.3. Release notes and signed binaries are published at bitcoincore.org, which the project treats as the authoritative source rather than third-party mirrors. Operations serving exchanges, mining pools or other high-availability infrastructure are better served by a staged rollout with testing before deployment.
The urgency is straightforward. With a full technical description now public, anyone can reproduce the attack against a node that has not been updated. The patch-first model kept the period of public, unpatched exposure short, but that period has begun.
For the market, the practical exposure runs through the intermediaries that depend on node uptime. An exchange running an affected node could see deposit crediting delayed, and a miner could burn computing power working from a stale view of the chain, though neither scenario appears in the advisory text itself. Neither outcome touches custody of funds or the ledger's integrity.
The broader signal is governance. As automated code analysis gets cheaper and more capable, the pool of vulnerabilities classed as easily discoverable will widen, and the interval between a fix and its public disclosure will keep shrinking. For node operators, staying current on releases is shifting from routine housekeeping to a security-critical function.
Once added, BigGo Finance appears first in Google Search Top Stories, so you get the broadest, most up-to-the-minute, and most comprehensive global financial news first.



