Yale proposes copyleft rules for AI trained on open-source code

Yale Researchers Propose Copyleft Licensing Rules for Generative AI

Researchers at Yale’s Digital Ethics Center have introduced a novel licensing framework that would require AI developers who train models on open-source code to make their architectures and training data publicly available, extending a decades-old principle from the free software movement into the era of generative AI.

A New License for a New Problem

The proposal, called the Contextual Copyleft AI License (CCAI), was published in the International Journal of Law and Information Technology and formally announced by Yale News on Monday. The framework treats generative AI models trained on free and open-source software as derivative works — a legal classification that would trigger reciprocal transparency obligations for developers who use such code.letsdatascience

“Our analysis showed that extending the copyleft concept to generative artificial intelligence has the potential to give open-source software developers meaningful control over how AI developers use their code,” said Grant Shanklin, a de Vries-Sherif Junior Fellow at the DEC and rising senior at Yale College, according to Yale News. “Importantly, it would incentivize the formation of a community committed to building AI tools aligned with the values of the free and open-source movement.”letsdatascience

The paper was co-authored by Shanklin, Emmie Hine, Claudio Novelli, Tyler Schroder, and Luciano Floridi, the founding director of Yale’s Digital Ethics Center.rts2

Extending Copyleft to AI

Traditional copyleft licenses, such as the GNU General Public License, require that derivative works of open-source software remain open. The CCAI extends this principle by classifying certain model artifacts — including architecture and training data — as derivatives of code inputs, attaching disclosure requirements to those artifacts.oup

The proposal arrives amid growing tension between proprietary AI model development and the norms of the open-source community. Companies and research groups have increasingly trained large models on public code repositories without reciprocating the transparency that open-source contributors expect.letsdatascience

Legal Questions Remain

The paper, published in Oxford’s International Journal of Law and Information Technology, evaluates the legal feasibility, policy justification, and risks of the approach. The authors argue that copyleft can serve core free and open-source software values when paired with responsible AI regulation. However, no finalized legal text has been released, and no court precedent currently enforces such a classification. Whether prominent open-source projects or downstream platforms adopt the framework — and whether it could survive legal challenge — remains to be seen.oup