Bitcoin Core Updates Vulnerability Disclosure Policy Due to AI

Bitcoin Core developers will expedite the release of patches and the disclosure of vulnerabilities that can be easily detected, including through simple AI queries. The policy change was adopted on October 22. Antoine Poinsot, the proposal’s author, explained that the revision was prompted by an increase in reports of issues that are easily rediscovered. Large language models have lowered the threshold for identifying certain classes of errors and can recognize fixes made without public notification. According to the developer, keeping such issues secret for extended periods is pointless and hinders the timely provision of protection to users. When reviewing a report, the team can classify a vulnerability as easily detectable. A simple query to a language model is one example, but this criterion is not limited to that.

First reported by ForkLog. This page is a short excerpt; the full text is theirs.