General Resolution: LLM Usage in Debian
7 hours ago
- Debian is holding a General Resolution (GR) to decide on the use of Large Language Models (LLMs) in contributions, with four competing proposals.
- Proposal A calls for a complete ban on LLM-assisted contributions to Debian, citing copyright, quality, community, and ethical concerns.
- Proposal B allows AI-assisted contributions if contributors meet conditions on legal compatibility, accountability, disclosure, and prior discussion.
- Proposal C requests that contributors avoid LLMs, mandates human-only drafting of messages, and allows project-level bans, treating violations as Code of Conduct issues.
- Proposal D accepts AI contributions for Debian-specific work, requiring DFSG compliance, submitter responsibility, disclosure, and caution with sensitive data.
- Key concerns across proposals include legal clarity of AI output, quality/reliability, impact on community and reviewer burnout, and unethical practices of AI companies (e.g., scraping, resource use).
- All proposals emphasize that the contributor remains fully accountable for any AI-assisted work they submit.