Canonical backs quest to translate mountains of C into safe Rust with AI
The Register
Canonical is funding a 3-year PhD to test if AI can turn big C codebases into safe Rust. It’s a research bet, not a bot takeover of Ubuntu.
Based on reporting by The Register — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Canonical is putting money behind a question the industry loves to hand-wave: can AI help turn huge C codebases into safe, maintainable Rust? Engineering vice president Jon Seager announced the co-funded, three-year PhD project on Ubuntu’s Discourse forum, and the work will run at the University of Bristol’s Programming Languages Research Group.
This is not Canonical asking bots to rewrite all of Ubuntu. The project is much narrower and much more academic: study whether an LLM can break down programs with hundreds of thousands of lines of C, then rewrite those pieces in Rust without wrecking the behavior along the way. That distinction matters. Canonical is funding a researcher, not buying a promise.
Seager’s pitch is careful about the limits of existing tools. Source-to-source translators can spit out Rust that compiles, he wrote, but they often drag C habits and unsafe operations straight through the conversion. The result still needs a lot of manual cleanup before a Rust maintainer would want to own it. The proposed research will look at whether a better pipeline can do more than produce awkward machine-shaped code.
The project also has a practical target list: snap-confine and AppArmor. That gives it a concrete shape, especially since AppArmor still has users in Debian and smaller distributions even as other enterprise Linux corners have leaned toward SELinux. And because the code in question is old and battle-tested, the project may surface the sort of behavior that lives only in years of fixes, patches and undocumented corner cases.
Canonical is co-sponsoring the work with UK Research and Innovation, and Seager will oversee it alongside Professor Meng Wang and Dr Cristina David. Three years is a normal PhD window in the UK, unless it isn’t, which seems to be the joke Canonical is quietly making at its own expense. The real value here is evidence. Either AI translation starts to look useful for legacy software, or everyone gets another data point on why this is harder than the hype crowd admits.
My take — AI-written commentary, not fact-checked reporting
This is the right kind of AI spending: narrow, boring, and measured in years instead of vibes. Canonical has at least noticed that legacy code is full of ghosts, and that rewriting can destroy the very behavior you wanted to preserve. More companies should fund research like this and fewer should act as if a prompt is a maintenance strategy.
Read more about this at: The Register