'The continued flood of AI reports has basically made the security list almost entirely unmanageable': Linus Torvalds laments how people are wasting the Linux team's time with LLMs

Linus Torvalds is well known for his blunt, no-nonsense approach to software development. In a recent post on the Linux mailing list archive regarding the latest kernel release candidate, he voiced significant frustration regarding the rising tide of low-effort bug reports generated by artificial intelligence.

The issue isn't the technology itself, but rather how it is being utilized by contributors. Instead of providing meaningful fixes, many users are flooding the development pipeline with automated messages that essentially offer no more than a notification that an LLM found a potential issue.

The Problem With AI-Generated Bug Reports

During his update, Torvalds noted that while new drivers—particularly GPU ones—make up roughly half of the kernel updates, other changes cover networking, core kernel, filesystems, and arch updates. However, he quickly pivoted to address the growing noise in the documentation and security sectors.

Torvalds specifically highlighted the chaos caused by uncurated automated submissions:

  • Massive Duplication: Different users are using identical tools to find the same issues.
  • Unmanageable Workloads: The sheer volume of reports is making the security list nearly impossible to oversee.
  • Lack of Value: Most reports consist of nothing more than "I used AI and it found this bug," without any accompanying context or solutions.

How to Add Real Value to the Linux Kernel

To be clear, Torvalds isn't anti-AI. He acknowledges that Large Language Models (LLMs) are excellent tools for engineers to offload drudgery or prototype ideas before committing code. The problem arises when anyone with access to a model can scan millions of lines of code and submit the findings without doing any legwork.

"If you found a bug using AI tools, the chances are somebody else found it too," Torvalds warned. He emphasized that for a submission to be worthwhile, contributors must go beyond simple detection.

To actually contribute effectively, Torvalds suggests:

  1. Read the documentation thoroughly.
  2. Create a patch to fix the identified issue.
  3. Provide actual value on top of what the AI discovered.

As the "flood of AI reports" continues, it is highly likely the Linux team will implement automated filters to weed out these low-effort submissions. If those filters end up being powered by AI themselves, it would be a fittingly ironic solution to the current bottleneck.