Foundation models are a new frontier of value creation in the digital platform economy. These technologies rely on the production and consumption of massive datasets that are monetised through consumer facing artificial intelligence (AI) products. However, the unlicensed use of these materials by the AI industry has provoked a legal and conceptual dispute. Media industries claim that the material in datasets is ‘content’ and subject to copyright law. AI industries, alternatively, are working to strategically reframe those materials as ‘data’, which is governed through regimes more congenial to the industry's business models, such as loosely enforced data protection and technologically secured trade secrets. This essay shows how AI copyright litigation, and the central question of data versus content, is mediating between different claims to the right to generate and justify value from datasets, as well as participating in the broader reformation of AI dataset markets.
Jake Goldenfein (Fri,) studied this question.