Developer questions whether AI coding agents should push open-source licensing toward AGPL copyleft protections

Julio Merino, who has contributed to open-source projects for almost 30 years and previously favored BSD-like licenses for their permissiveness, is reconsidering that stance after two months of using AI coding agents for real development work.

AI agents can closely mimic a project's architecture and coding style with zero effort, making it trivially easy for anyone to fork existing free software, customize it via agents, and redistribute modified versions without the significant time investment and deliberation that modification used to require.

Merino worries that slop forks (low-quality but functional AI-generated variants) may accumulate features faster than upstream maintainers can review them, potentially rendering the original project irrelevant and creating pressure on maintainers to keep pace or lose community interest.