Legal Scholar Argues Copyright Law Cannot Absorb AI Training

Legal scholar David Simon argues in a new Substack analysis that historical copyright adaptation mechanisms are insufficient for addressing generative AI training disputes. Simon contends that previous legal strategies for absorbing technological disruption will likely fail when applied to large language models ingesting protected works at scale. The article suggests current fair use precedents may not protect AI developers from infringement claims due to the unique nature of machine learning extraction. This analysis arrives as multiple federal courts await rulings on whether unauthorized training constitutes transformative use or market substitution. Simon’s position challenges industry assumptions that existing copyright flexibility can accommodate AI without legislative intervention. Legal experts note this perspective aligns with growing judicial skepticism toward broad tech exceptionalism arguments in intellectual property cases. The piece contributes to ongoing debates regarding potential statutory licensing schemes for AI training data.