Codeberg e. V. members have voted to prohibit the use of platform data for training large language models and to restrict the hosting of automated, AI-generated software projects.

Key Points

• Members passed two motions to protect the free and open-source software ecosystem from the impacts of generative artificial intelligence.
• Codeberg officially prohibits the use of its user and project data for training large language models.
• New terms of use will discourage "vibe-coded" projects, defined as resource-heavy, unmaintained software created primarily by AI agents.
• The organization cited rising hardware costs, server strain from web crawlers, and the erosion of human-centric collaboration as primary drivers for the policy change.
• Enforcement will focus on high-resource, autonomous projects rather than individual experiments or human-led projects that utilize AI as a secondary tool.

Why it Matters

This decision highlights a growing tension between open-source communities and the resource-intensive demands of generative AI developers. By prioritizing human collaboration over automated output, Codeberg aims to preserve its infrastructure and maintain the integrity of its software commons against the rising costs and social challenges posed by AI-generated content.