European Technology & Digital Sovereignty | 8th November 2026
Artificial intelligence has demonstrated a major advance in cybersecurity after researchers at OpenAI used an automated system to identify and exploit hundreds of vulnerabilities in the Linux kernel.
Published on 4 November, the research, titled A Thousand Fewer Paths to Root, examined security weaknesses in the open-source software underpinning Linux operating systems, servers and much of the world's digital infrastructure.
The system identified 2,308 potential security flaws, reproduced 1,724 vulnerabilities and demonstrated working exploits for 948 distinct weaknesses capable of allowing a local user to gain administrator-level access.
The research cost approximately $1.2 million in computing credits, demonstrating how AI could dramatically accelerate vulnerability discovery.
OpenAI's researchers also generated proposed security patches, with human experts reviewing the findings before submitting them to Linux maintainers.
At the conclusion of the investigation, 61 vulnerabilities had already been patched upstream, while another 43 had proposed fixes awaiting integration. These figures do not mean that all 948 vulnerabilities remain exploitable on updated systems.
The discovery highlights both the benefits and risks of AI-assisted cybersecurity. While defenders can identify weaknesses more quickly, increasingly capable AI systems could also make vulnerability discovery easier for attackers.
For Britain and Europe, where Linux supports public services, businesses and independent digital infrastructure, the research reinforces the importance of maintaining secure, regularly updated open-source systems.