Washington Journal of Law, Technology & Arts


Machine learning services ingest customer data in order to provide refined, customized services. Machine learning algorithms are increasingly prominent in multiple sectors within the software-as-a-service industry including online advertising, health diagnostics, and travel. However, very little has been written on the rights a company utilizing machine learning needs to obtain in order to use customer data to improve its own products or services. Machine learning encompasses multiple types of data use and analysis, including (a) supervised machine learning algorithms, which take specific data provided in a tagged and classified format to deliver specific predictable output; and (b) unsupervised machine learning algorithms, where untagged data is processed in order to look for patterns and correlations without a specified output. This Article introduces the reader to the types of data use involved in various machine learning models, the level of data retention normally required for each model, and the risks of using personal information or re-identifiable data in connection with machine learning. The paper also discusses the type of license a commercial provider and consumer would need to enter into for various types of machine learning software. Finally, the paper proposes best practices for ensuring adequate rights are obtained through legal agreements so that machines may self-improve and innovate.

First Page