The topic of API copyright was top of mind for most of the time between 2010 and 2020 due to the Oracle vs Google API copyright case. That was settled in early 2021 with APIs being copyrightable now the law of the land in the United States. Honestly it really didn’t change much, as most companies saw the naming and ordering of their APIs as something that is proprietary and their secret sauce despite convincing hundreds or thousands of developers to bake it into their application. This left us believing that API copyright doesn’t really matter, and that the API copyright debate would come down to who would litigate in defense of their APIs, which is something we see beginning to unfold in the artificial intelligence space in a couple of dimensions of the often obfuscated realm.
- Microsoft and OpenAI Exclusive Partnership
- Deepseek Improperly Using the OpenAI API
- OpenAI OpenAPI License
- OpenAI Copyright Disputes)
How does “The OpenAI API is exclusive to Azure, runs on Azure and is also available through the Azure OpenAI Service. This agreement means customers benefit from having access to leading models on Microsoft platforms and direct from OpenAI.” impact the API copyright debate, with a MIT “code” license and not a copyright license applied to OpenAPI. I am not a lawyer, so I won’t even pretend to understand this stuff, but I do have a nuanced understanding of the separation between code and interfaces, and how we describe these interfaces in machine-readable and copyrightable ways. I feel like all of this will be blown out of the water not because of API copyright or licensing, but because of the lack of respect for copyright in general, unless it is your IP. Regardless, the ongoing discussion makes for an interesting opportunity to keep shining a light on why APIs matter in this whole AI drama.
