The short answer is, yes.
As retail and fashion companies increasingly use artificial intelligence (AI) and machine learning technologies to perform predictive consumer analysis, solve apparel fit issues, create virtual and physical fashion and other designs, these technologies must be used in a manner that does not infringe on the intellectual property rights of third parties.
So, what is artificial intelligence?
Broadly speaking, artificial intelligence involves using computers to emulate the manner in which the human brain performs certain computations. One area of AI, commonly referred to as machine learning using neural networks, involves extrapolating patterns from large quantities of data and using those patterns to generate actionable information or output data — e.g., consumer purchasing trends, apparel designs, etc. Over time, the AI program refines its underlying algorithm, which results in increasingly higher quality output data. This process is commonly referred to as training a neural network.
What do artificial intelligence and copyright have in common?
As mentioned above, training a neural network involves extrapolating patterns from large quantities of data and using those patterns to generate actionable information or output data. Such large quantities of data, however, may contain copyright protected images or the like — for example, copyright protected photographs downloaded from the internet. And using these copyright protected images as input data to train neural networks can, in certain circumstances, constitute copyright infringement.
For example, the mere downloading and temporary use of copyrighted materials only for the limited purpose of training a neural network, may constitute infringement. In particular, some U.S. courts consider temporarily downloaded copies of copyright protected material to be infringement, whereas other courts require more than fleeting downloaded copies to find infringement.
Additionally, copyright infringement may occur when the output model (e.g., AI-produced virtual design) that results from the above-mentioned machine learning is substantially similar to the copyright protected materials used as input data during the machine learning process. This may occur when homogenous or over-fitted input data sets are used to train one's neural network.
Given the massive amount of digital data available to retail and fashion companies, and the equally massive opportunities for these companies to use AI systems and machine learning to create efficiencies, predict consumer behavior and so on, special care should be taken to avoid unwanted allegations of intellectual property infringement.
One obvious way to prevent copyright infringement is to avoid the unauthorized use of copyright protected materials.
Companies also can advise their engineers or designers against completely homogenous or over-fitted input data sets. The less copyrighted material, both qualitatively and quantitatively in relation to the whole, the neural network is using in machine learning, the greater the chance that the use of the copyrighted material is permissible.
And finally, if the use of copyrighted materials cannot be avoided during neural network training, it may be excused under a legal doctrine called "fair use." The fair use doctrine promotes freedom of expression by permitting the unlicensed use of copyright-protected works in certain circumstances based on four factors: (1) Purpose and character of the use, including whether the use is of a commercial nature or is for nonprofit educational purposes; (2) Nature of the copyrighted work — i.e., the degree to which the work that was used relates to copyright's purpose of encouraging creating expression; (3) Amount and substantiality of the portion used in relation to the copyrighted work as a whole; and (4) Effect of the use upon the potential market for or value of the copyrighted work.
This article was originally published in Retail TouchPoints on November 5, 2021. It is co-authored by partner Gina Bibby and intellectual property / technology summer 2021 intern Erica Barseghian.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.