Lin Yang and Taorui Guan: “From Safe Harbours to AI Harbours: Reimagining DMCA Immunity for the Generative AI Era

Lin, Yang, and Taorui Guan. “From Safe Harbours to Ai Harbours: Reimagining DMCA Immunity for the Generative Ai Era.” Journal of Intellectual Property Law and Practice, Oxford Academic, 4 Aug. 2025, academic.oup.com/jiplp/article/20/9/605/8221820. Accessed 29 Oct. 2025.

This article addresses many of the same issues we’ve discussed this week already on the complications that are being faced in the aftershocks of the Anthropic decision. Yang and Guan in this piece seek to suggest a new clause built off of the DMCA Section 512, where immunity to claims is specifically tied to the disclosure of provenance, transparency of data suppliers, and more. This hypothetical measure would require that the authors of the works that ultimately are used in LLM development are compensated, and works that stem from the use of generative AI are watermarked respectively, and give reference to the works that it derives content from.

This piece suggests a risk-based approach in addressing black box AI systems, where regulations would be prioritized based on the number of users, sensitivity of data used in training, and any history of infractions against such legal regulation. Yang and Guan’s work would provide what I could examine to be a potentially applicable fix, at least in part, to the issue on the opacity of generative AI models.