Canonical funds three-year PhD to test AI-driven C-to-Rust code translation

Canonical has committed funding for a three-year PhD program at the University of Bristol's Programming Languages Research Group, aiming to assess the feasibility of using large language models to translate extensive C codebases into safe, maintainable Rust. The initiative, announced by engineering vice-president Jon Seager, is jointly financed by UK Research and Innovation, a public body under the Department for Business, Innovation, Science and Trade.

Researchers will evaluate whether AI can decompose hundreds of thousands of lines of C into smaller units and rewrite them without relying on unsafe constructs, targeting components like snap-confine and AppArmor. Prior Ubuntu releases have already incorporated hand-written Rust replacements for sudo and core utilities, but this project seeks an automated approach. Seager acknowledges the challenges of preserving undocumented knowledge embedded in legacy code and hopes the study will provide concrete evidence of AI's utility in software modernization. The outcome could influence how open-source projects handle security-critical components in the future.