Protecting our FLOSS commons from LLMs
blog.codeberg.org
July 23, 2026
10 min read
🔥🔥🔥🔥🔥
49/100
Summary
Codeberg e. V. members voted on two motions regarding artificial intelligence and Large Language Models (LLMs), affirming a commitment to not use member data for LLM training. The organization expressed concerns that LLMs pose a risk to the free/libre software ecosystem.
Key Takeaways
- Codeberg e. V. voted to prohibit the use of user data for training Large Language Models (LLMs) and believes these technologies threaten the free and open-source software ecosystem.
- The organization implemented a change to its terms of use to ban 'vibe-coded projects' following a vote with significant member participation.
- LLMs impose hidden costs on society, including increased hardware prices, energy consumption, and environmental damage, affecting both users and non-users.
- Codeberg's infrastructure faces strain from excessive web crawling by companies attempting to gather code for LLM training, which impacts service quality and requires defensive measures.
Community Sentiment
Positives
- Some see Codeberg's democratic decision-making as a necessary step, even if it leads them to move their projects elsewhere — it's about principles over convenience.
- Commenters appreciate the intention behind Codeberg's new policy, viewing it as a potential safeguard against AI-generated code that lacks clear ownership.
Concerns
- Critics argue that the new TOS banning AI-generated code is an overreaction, raising concerns about stifling innovation and the natural evolution of software development.
- Skeptics point out that the decision may alienate developers who rely on LLMs, potentially turning Codeberg into a niche platform disconnected from modern coding practices.
- There's a palpable frustration regarding how democratic processes can suppress minority voices, with some likening the voting outcome to moralistic pile-ons rather than genuine community consensus.
