Intellectual Property Reform
Adapted copyright, patent, and licensing frameworks that clarify rights over AI training data, establish compensation mechanisms for creators whose works train AI systems, and determine the ownership status of AI-generated outputs.
What it is:
Intellectual property reform for AI addresses the legal frameworks governing both the inputs to and outputs from AI systems. On the input side, the central question is whether and how creators should be compensated when their copyrighted works (text, images, music, code) are used as training data for AI models. On the output side, the question is whether AI-generated content should receive intellectual property protection, and if so, who owns it: the developer of the model, the user who prompted it, or no one.
IP frameworks could determine whether productivity gains from AI accrue primarily to model developers or are shared with the creative labor force whose works made those models possible.Reform proposals range from expanding "fair use" exceptions to permit training (prioritizing AI development and access), to mandatory licensing schemes modeled on the music industry's collective rights organizations (prioritizing creator compensation), to transparency requirements that would at minimum enable creators to know whether their works were used.
The challenge:
The main challenge is that the legal landscape is fragmented and internationally inconsistent. Courts in different jurisdictions are reaching contradictory conclusions on whether AI training constitutes fair use, creating uncertainty for both developers and creators. Any licensing scheme must solve the aggregation problem: millions of individual creators each hold small claims that are impractical to negotiate individually but collectively represent enormous value. Opt-out mechanisms place the burden on creators to monitor and enforce their rights across rapidly proliferating AI systems, while opt-in mechanisms could make training data prohibitively difficult to assemble.
Recommended Reading:
Real-world precedents:
The music industry's collective licensing infrastructure — where organizations like ASCAP, BMI, and SESAC pool rights to negotiate blanket licenses with radio stations and streaming services — offers a model for aggregating fragmented copyright claims at scale.
In February 2025, Thomson Reuters v. Ross Intelligence became the first case rejecting fair use for AI training, finding that using Westlaw headnotes to train a competing legal research tool was not transformative. However, in June 2025, Kadrey v. Meta and Bartz v. Anthropic found, on the specific facts before them, that AI training was highly transformative fair use.
The EU's Digital Single Market Directive established opt-out text and data mining exceptions, while the EU AI Act requires general-purpose AI model providers to publish training data summaries that enable rights holders to exercise their rights.