The author argues that generative AI is dismantling the established equilibrium between software copyright and open-source culture. Historically, the evolution of "copyleft" frameworks—such as the GNU General Public License (GPL), Mozilla Public License (MPL), and Creative Commons ShareAlike (CC-SA)—allowed for a collaborative ecosystem that built the modern internet and cloud infrastructure. However, Large Language Models (LLMs) now ingest vast amounts of online data without adhering to these original licensing obligations, effectively breaking the social contract of digital openness. This shift transforms the act of sharing knowledge from a communal gift into a personal liability for creators. The author highlights several technical and social risks: the proliferation of "slop code" and AI-generated pull requests on platforms like GitHub, which increases the triage burden for maintainers; the potential for AI to more easily identify vulnerabilities in public codebases; and the risk of developers inadvertently incorporating stolen work or malicious libraries into their projects. Ultimately, the author warns that the erosion of enforceable licensing and the unchecked consumption of derivative works by AI could lead to a "digital dark age," where consolidated power replaces the empowering, decentralized progress achieved over the last 35 years of open software development.