Anthropic researcher Nicholas Carlini utilized Claude Code to discover several security vulnerabilities within the Linux kernel, including one that had remained undetected for 23 years. Carlini was surprised by the AI's effectiveness, noting that he had personally struggled to find such complex bugs. The process involved directing Claude Code to analyze the kernel's source code with minimal oversight, prompting it to search for vulnerabilities within specific files. AI
Summary written by gemini-2.5-flash-lite from 1 source. How we write summaries →
IMPACT Demonstrates AI's potential in uncovering complex software vulnerabilities, potentially accelerating security auditing.
RANK_REASON AI model used to find security vulnerabilities in open-source software. [lever_c_demoted from research: ic=1 ai=1.0]