The Fedora Project just made a big decision about how contributors can use Artificial Intelligence (AI). After weeks of community discussion and gathering feedback, the Fedora Council officially approved a new AI contribution policy. This new rule sets the guidelines for contributions helped by Large Language Models (LLMs) and other generative AI tools.
This policy allows contributions that use AI, but only if people follow clear standards. Moving forward, the developers can use tools like ChatGPT and other LLMs—but only under strict rules of accountability and transparency.
Table of Contents
Why Fedora Adopted the AI Policy
This decision came after a large, collaborative discussion with the community.
Many contributors already use AI tools, like Copilot, to speed up their work. AI tools help developers write boring tests or generate boilerplate files.
Furthermore, AI is very good at following the project's specific code style. For example, a tool can quickly create about 60 lines of code based on a prompt.
This saves valuable time that a contributor would normally spend looking up documentation or flags. Thus, developers often view AI as a valuable helper for menial tasks.
However, the use of AI stirred up intense debate. Many people worried about legal issues and code quality. The Council, therefore, set three strict principles for the new policy.
The Three Key Rules for AI-Assisted Code
The new policy uses simple, strong language, like "MUST", "SHOULD" and "MUST NOT", to define expectations.
1. Accountability is Non-Negotiable (MUST)
Every human contributor MUST take full responsibility for their submission. Contributing to Fedora means you vouch for the quality, license compliance, and utility of the code.
The policy makes this requirement absolutely clear: The contributor is always the author and is fully accountable for the entire contribution, whether a human wrote it or an AI assisted it.
This rule ensures that the legal and quality burden stays with the human submitting the patch.
2. Transparency is Required (MUST and SHOULD)
Contributors MUST disclose AI tool use when they take a significant part of the contribution from a tool without changes. They SHOULD also disclose other uses of AI when this information might be helpful.
This disclosure rule fosters transparency. Transparency helps the community evaluate the AI tools' impact.
For code changes tracked in Git, the recommended way to disclose is by using an Assisted-by: commit message trailer. For example, a contributor might write Assisted-by: generic LLM chatbot.
Conversely, minor assistance like fixing grammar or spelling does not require disclosure.
3. AI Cannot Judge Humans (MUST NOT)
While AI tools can help human reviewers by offering suggestions or analysis, the AI MUST NOT act as the sole or final decision-maker. Furthermore, Fedora MUST NOT use AI to evaluate a person's standing in the community, such as for leadership roles or funding.
This rule makes certain that human judgment remains central to the project’s governance.
What Happens Next?
The Fedora Council knows this policy will need updates in the future. AI technology changes rapidly, so the policy will utilize the Policy Change Policy when revisions become necessary.
The Council will post the final outcome to the discussion thread, the mailing list, and the community blog. Then, the agreed policy will officially join the Council Policies page.
The policy specifically excludes large-scale initiatives that might dramatically change the project; such big changes need separate discussion with the Fedora Council.
