A recently-filed federal court complaint tests the enforceability of restrictive terms in a data license against the use of licensed data for generative AI purposes. The outcome of this case may turn on interpreting broad terms such as training, internal research, distribution and publication.
UPDATE: On December 18, 2025, the court denied defendant Alexi Technologies Inc.’s (“Alexi”) motion for a temporary restraining order (TRO) based on Alexi’s contention that it would suffer irreparable harm without access to Fastcase’s updated data and lose revenue and customers and suffer reputational damage. In a brief order, the court found that Alexi’s harm is “too speculative and is not sufficiently corroborated to merit emergency relief at this time.”
On November 26, 2025, legal research platform Fastcase, Inc. (“Fastcase”) sued legal AI company Alexi Technologies Inc. (“Alexi”) in a District of Columbia district court, alleging that Alexi used licensed “Fastcase Data” (i.e., a sophisticated, extensively tagged legal research database) to train and power a commercial AI legal research product in violation of a 2021 data license agreement.(Fastcase, Inc. v. Alexi Technologies Inc., No. 25-04159 (D.D.C. Filed Nov. 26, 2025)).
The complaint is particularly interesting as it involves a 2021 agreement – entered into before generative AI was a generally known technology – and thus presents the age-old question of interpreting the terms of an agreement to a technology not known at the time the agreement was entered into.
Regardless of the outcome of this case, the complaint highlights the importance of precise license drafting – for both licensors and licensees – in the age of AI. Parties should focus on well-defined understandings of how data can be used, shared and commercialized at a time where every company is seeking to leverage AI. As this complaint illustrates, vague terms can lead to uncertainty and legal disputes.[1]
Of course, we do not know if this litigation will proceed to summary judgment or trial, but if it does, we may have an opportunity to see how a court will interpret and potentially enforce broad and general restrictions in a data license (e.g., no publication, distribution, commercial use, competitive use, use for internal research purposes only) in the context of generative artificial intelligence.[2]