Torvalds: AI bug hunters clog Linux security
Linus Torvalds has called the Linux security mailing list "unmanageable" due to a flood of duplicate bug reports from AI tools like Claude Mythos. New rules now classify AI-detected bugs as "public by default" to reduce administrative churn and force researchers to provide verified reproducers or patches.
The drive-by automated research era is hitting its first major bottleneck as maintainer burnout from AI-generated noise reaches a breaking point. Torvalds argues that bugs found by commodity AI are not secrets, and reporting them publicly prevents redundant triaging of identical reports. This noise has turned kernel maintenance into "make-believe work," shifting focus away from actual development. The community is now raising the bar, treating automated reports without a verified reproducer or a proposed patch as low-value noise. Meanwhile, maintainers like Greg Kroah-Hartman are successfully using custom AI tools like "Clanker T1000" to fix stable-branch bugs.
DISCOVERED
13h ago
2026-05-18
PUBLISHED
17h ago
2026-05-18
RELEVANCE
AUTHOR
jonbaer