Bitcoin Core Changes Security Rules As AI Makes Finding Code Bugs Dangerously Easy

TL;DR

Before you orderOffer from Amazon

Get hardware and tech essentials delivered free with Prime

• Fast, free delivery on millions of items
• Prime Video, Amazon Music and more included
• Member-only deals all year

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Bitcoin Core maintainers have changed the project’s security policy so vulnerabilities judged easy to rediscover with an AI prompt can be patched and disclosed quickly, rather than kept under a long embargo. The change may reduce the time users wait for a fix, but maintainers say those releases could receive less testing before they reach users.

Bitcoin Core maintainers have changed the project’s vulnerability-disclosure policy to speed up fixes for bugs that can be rediscovered with a simple prompt to an AI model. Under the policy merged Thursday, those vulnerabilities can be patched across supported versions and disclosed with a security advisory as soon as possible, rather than held under an embargo that could last months.

The revised policy lets maintainers classify a reported vulnerability as easy to discover during triage. The practical test described in the policy is whether it “can be found with a simple prompt to an LLM,” or large language model. If it meets that test, maintainers plan to backport a fix to every version of Bitcoin Core still supported and issue a point release with an advisory.

Previously, Bitcoin Core timed disclosure according to severity. Low-severity issues were disclosed two weeks after the next major release, while medium- and high-severity vulnerabilities could remain under embargo until two weeks after the last affected version reached end of life. The aim was to give node operators time to install a fix before public details could help attackers.

The new approach trades some of that quiet upgrade time for faster access to a repair. Antoine Poinsot, a Chaincode Labs developer who proposed the change, said fixes for these issues would have less “bake time” for testing before they reach users. No vulnerability has yet been identified as subject to the new rule, according to the report.

Faster Fixes, Shorter Testing Windows

Bitcoin Core is the software used by most of the Bitcoin network, including the nodes that validate and relay transactions. A vulnerability in the software can therefore matter to node operators and to the network’s broader security. The policy change reflects a practical concern: if AI tools make a flaw easy to find again, keeping its existence secret may no longer give users much protection.

At the same time, a faster release can leave operators with less time to assess and test a fix before installing it. Poinsot acknowledged that the fixes would have less testing time. That means the change is not simply a decision to disclose more quickly; it shifts the balance between limiting information available to attackers and getting a correction to users sooner.

The policy also makes disclosure itself part of the risk calculation. Some developers argued that Core should retain the option to release a fix without publishing an advisory, or to delay details in individual cases. Poinsot rejected that approach, saying an alert without details could prompt unnecessary, rushed migrations by users who were not affected.

AI Reports Prompt Policy Review

The change follows a reported rise in security findings generated or assisted by AI tools. In his proposal, Poinsot said maintainers had seen more reports about vulnerabilities that could be easily rediscovered, and attributed the trend in part to large language models lowering the difficulty of finding some classes of issues or recognizing covert fixes.

One illustration of the broader pressure came in late July, when a volunteer group of 16 developers calling itself the Bitcoin Red Team tested AI models across 390 Bitcoin-related projects in about a day. The group recorded 4,962 potential findings, including 85 it classified as critical. Those figures describe the group’s reported results; they do not establish that every finding was independently confirmed or exploitable.

Some reported vulnerabilities had immediate consequences outside Bitcoin Core. The source report said a BTCPay Server flaw disclosed privately to the project was exploited on live merchant servers within days, draining funds from Lightning nodes used for payments. Core Lightning also advised operators to disconnect after AI-assisted reports to its small team proved to be real. Separately, a Coldcard hardware-wallet firmware flaw that had been present since 2021 was exploited on July 30, causing losses across four sweeps; the report said the incident did not depend on AI to find the bug.

“We recently experienced an increase in the number of reports for security vulnerabilities that can be easily (re-)discovered.”

— Antoine Poinsot, Chaincode Labs developer and author of the proposal

How the Rule Will Work in Practice

No bug has yet been flagged under the new rule, so its effects on release timing, user behavior and security outcomes are not yet known. The policy’s simple-prompt test gives maintainers a working criterion, but the report does not specify a formal review process for deciding whether a particular vulnerability qualifies.

It also remains unclear how much testing time a rapid point release will receive in practice, or whether maintainers will use case-by-case judgment when disclosure could create risks for users. The policy change does not establish that AI can find every vulnerability, nor that every AI-generated report is valid. The Bitcoin Red Team’s reported findings were potential flaws, and the source material does not say how many were independently verified.

First Cases Will Test the Policy

The next practical test will come when maintainers triage a vulnerability that they judge easy to rediscover with an AI prompt. They will need to determine whether the issue meets that standard, prepare fixes for supported versions, and decide when the point release and advisory are ready. The project has not announced a specific first case or a timetable for one.

Node operators and other users will need to follow Bitcoin Core’s release notices and security advisories to learn whether a new fix affects their software. Until the policy is applied to a real vulnerability, the balance between quicker patches and shorter testing windows remains a stated tradeoff rather than a measured result.

Key Questions

What did Bitcoin Core change?

It added a process for quickly patching and disclosing vulnerabilities that maintainers judge can be found with a simple prompt to an AI model. The fix is intended to reach all supported versions in a point release, accompanied by a security advisory.

Why is Bitcoin Core shortening some disclosure delays?

Maintainers say AI tools have made some flaws easier to rediscover, reducing the protection an extended embargo can provide while delaying a fixed release. The new approach aims to get patches to users sooner in those cases.

Does the new policy mean AI found a Bitcoin Core bug?

No. The source report says no vulnerability has yet been flagged under the new rule. The policy responds to a broader increase in AI-assisted security reports and the possibility that some flaws are easy to reproduce with a prompt.

What is the main tradeoff?

Users may receive a fix sooner, but maintainers say the release could have less time for testing before it reaches them. How that affects real releases and users is not yet known.

Does the Bitcoin Red Team tally mean all 4,962 findings were confirmed bugs?

No. The group reported 4,962 potential findings across 390 projects, including 85 it classified as critical. The source material does not say that every finding was independently confirmed or exploitable.

Source: rss