The Linux creator distinguishes between using AI to find bugs and using it to write core system code, warning against vibe coding in critical infrastructure.
Key points
- Linus Torvalds supports using AI tools to identify bugs in existing code.
- He explicitly rejects using AI for writing new Linux kernel development code.
- Torvalds describes vibe coding as enjoyable for general programming but unsuitable for kernel work.
Linus Torvalds has clarified his stance on artificial intelligence in software development, drawing a sharp line between debugging and coding. The Linux creator says AI is useful for finding errors but dangerous for writing the kernel.
What happened
Linus Torvalds recently shared his specific usage patterns for AI coding assistants in interviews reported by ZDNET Security. He stated that he uses AI to handle tasks where he lacks skill or efficiency, particularly in finding bugs. Torvalds described this approach as a way to leverage the technology for its strengths while avoiding its weaknesses.
ZDNET reported that Torvalds called vibe coding a wonderful way to find joy in programming for general users. However, he firmly rejected the idea of using this method for running Linux development. He emphasized that the kernel requires a level of precision and control that current AI tools cannot guarantee.
The Linux creator made it clear that while AI can help locate issues in code, it should not generate the code itself for critical systems. This distinction separates casual programming from the rigorous demands of operating system maintenance.
Why it matters
The Linux kernel underpins much of the world's critical infrastructure, including servers, cloud platforms, and Android devices. Any reduction in code quality or introduction of subtle bugs could have widespread security and stability consequences. Torvalds' stance highlights the ongoing tension between developer productivity and system reliability.
Security teams and developers must recognize that AI-generated code may contain hidden vulnerabilities or logic errors. Relying on AI for core system components without rigorous human review increases the risk of introducing flaws that are difficult to detect. Torvalds' warning serves as a cautionary tale for organizations adopting AI tools in high-stakes environments.
The fix
- Restrict AI coding assistants to debugging and documentation tasks only.
- Require human review for all code generated by AI tools.
- Prohibit AI from writing new code in critical system components.
- Establish clear policies on AI usage in development workflows.
What to do and how to stay safe: Linux
- Audit your development team's use of AI coding assistants to ensure they are not writing core system code.
- Implement mandatory code reviews for any code generated or significantly modified by AI tools.
- Train developers on the limitations of AI in identifying subtle logic errors and security vulnerabilities.
- Monitor vendor updates for AI tools to stay informed about new features and potential risks.
Step-by-step guide: Social Engineering Risks and Protection for Small Businesses
General security guidance from the Crypt Weekly newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Can I use AI to write Linux kernel code?
Linus Torvalds explicitly rejects using AI for writing Linux kernel development code, though he approves it for finding bugs.
What is vibe coding?
Vibe coding is a term Torvalds used to describe a relaxed, AI-assisted approach to programming, which he finds enjoyable but unsuitable for kernel work.
Why does Torvalds allow AI for debugging?
Torvalds uses AI to handle tasks he is bad at, finding it an efficient way to locate bugs without compromising code quality.
