Codeberg Bans AI Code: What the 358-144 Vote Means

Codeberg voted to ban projects “mostly” written by generative AI tools. Result: 358 to 144 (71% in favor), roughly 50% member turnout. The ban is already live in the Terms of Use.

Updated Terms of Use: “You must not share projects that mostly consist of code written by ‘generative AI’-tools (including services such as Claude, OpenAI Codex).”

Two resolutions passed at Codeberg’s annual assembly on July 22, 2026. The first bans AI-heavy repos. The second is a binding commitment that Codeberg will never use project or user data to train LLMs. Both passed decisively.

What’s allowed: ordinary projects where AI contributes occasionally, small experiments, established communities that accept some LLM-assisted contributions. What’s not: repos where AI wrote the majority of the code, regardless of human review afterward.

The “mostly” problem

Codeberg has not defined a percentage threshold for “mostly.” There is no detection methodology. Enforcement will be complaint-driven by volunteers; enforcement details are still to be worked out. Critics including Armin Ronacher (Flask; Earendil) called it “a very bad move” (per The Register).

Why Codeberg did this

Cost: SSD and CI/CD costs rising; AI crawlers and AI-generated repos consume storage/CI budgets donations barely cover. Copyright: unresolved legal status of AI-generated code and “license laundering” risk. Community trust: “The widespread use of LLMs in FLOSS is becoming a multidimensional attack on the trust between contributors.” Ghost projects with heavy CI and no maintainers are cited.

Broader context

Debian’s General Resolution on LLM usage is still debating competing proposals. Irony: Zig left GitHub for Codeberg in 2025 over AI concerns; Codeberg now enforces that skepticism as policy. Alternatives for AI-heavy OSS: self-host Forgejo, Gitea, or stay on GitHub/GitLab.