Jason Brooks, a member of the Managing Council of Fedora, has initiated discussions regarding the draft version of rules regulating the use of AI tools for developing Fedora Linux. The proposal calls for a two-week period of discussion and amendments to the rules before members of the Council vote on the approval of the new requirements.
AI tools are seen as having the potential to enhance the platform, but concerns exist regarding confidentiality, safety, ethics, and quality. The proposed rules do not ban the use of AI assistants but emphasize the importance of human oversight and developer accountability. The following rules have been suggested:
- Developers using AI tools for changes in Fedora must take responsibility for the changes, study, check, and test them. The content generated by AI should be viewed as a recommendation, not the final code or text. Submitting unverified or low-quality AI content is deemed unacceptable.
- When submitting changes made with AI tools, the note to the Committee or Pull Requests must include the “Assisted-By” tag, specifying the name of the AI assistant.
- The use of AI tools is encouraged for overcoming language barriers and explaining thoughts in communication.
- When reviewing others’ changes, limited use of AI tools is recommended. AI can aid in forming reviews but cannot fully automate the process. The final decision on approving or rejecting changes should be made by a person.
- In project management, AI tools can automate routine tasks like spam filtering and note creation. However, their use is deemed inappropriate for issues like code of conduct violations, funding requests, leadership nominations, and conference report preparation.
- Proposed distribution of AI capabilities, including sending data to external servers, should be opt-in and activated by default only after informing and obtaining consent from the user. AI tools can be beneficial for tasks such as translation, transcription, and speech synthesis for people with disabilities.
- Packages with tools and frameworks necessary for AI research and development should be created.
- Data generated by the project can be used for AI models under licenses and proper attribution. Aggressive data scraping that burdens the infrastructure is prohibited. For efficient data access, contact the infrastructure team.
/Reports, release notes, official announcements.