systemd 262-rc2 Adds an AI Canary for Detecting Unreviewed AI/LLM Code Contributions
systemd 262-rc2 adds an intentionally conspicuous AGENTS.md instruction that asks an AI agent to insert a marker into README.md; humans must remove it after reviewing the contribution. The canary is meant to expose code that reached a pull request without human inspection, alongside ordinary rc2 fixes and hardware-database updates. It is a blunt but concrete example of turning agent behavior into a review signal rather than trusting a contributor’s declaration.
The 15-response forum thread largely centered on the canary; one commenter noted that the release had other interesting changes that received less attention. The mechanism is detection, not proof that every line was AI-generated.