BehindLines

Ubuntu: security fixes written by AI

· BehindLines newsroom

Canonical, the company behind Ubuntu, is now offering Ubuntu Pro subscribers early access to security fixes written by artificial intelligence. According to The Register and OMG! Ubuntu, they are distributed in a new optional repository named Edge.

A multi-agent system called Surge

According to The Register, Pierre Guillemin of Canonical presented the project at the recent Ubuntu Summit. Surge is described as a multi-agent orchestration system for fixing vulnerabilities and adapting existing fixes to Ubuntu versions. One group of language models triages new flaws and looks for fixes already published, then several competing models port them to the affected versions and launch automatic tests. Another group, tasked with judging, evaluates the proposals and seeks a consensus. In an edge case, with low confidence or a deadlock, the matter is passed up to Canonical's security team, according to the outlet. Surge only handles userspace software, not the kernel, both sites specify.

A repository for volunteers

According to OMG! Ubuntu, the machine-generated fixes are not switched on by default: subscribers must choose to enable the Edge repository, and each update can be picked by hand. Individuals can use Ubuntu Pro at no cost on at most five devices, which allows researchers and advanced users to test these fixes, the site says.

Canonical states that Surge's fixes will not be pushed automatically to stable releases. Updates delivered through the usual channels remain validated by humans, according to The Register and OMG! Ubuntu. The company explains that this human work can take longer than some customers are able to wait; Edge is aimed at those who would rather not hold off.

A context of rising flaws

The Register reports that Pierre Guillemin said October 2026 had counted almost as many catalogued vulnerabilities as the whole of 2025. OMG! Ubuntu adds that Canonical's security team has already published twice as many fixes as last year, in the face of a rise in reports of flaws found with the help of language models and of attacks carried out by AI agents. According to The Register, Pierre Guillemin sums up the idea as "fighting AI with AI", an approach the site considers bold.

Why it matters

Faced with this rise in flaws, Edge lets volunteers obtain fixes before full testing is finished. Canonical says it wants to validate their stability this way before promoting them to Ubuntu's LTS releases, according to OMG! Ubuntu. For now this is an evaluation phase: updates delivered through the usual channels remain validated by humans.

Sources