Ask Dr. Copyright

Dear Doc:

I have read your columns about how patent law is now mostly unable to protect inventions that incorporate artificial intelligence and machine learning computer applications. So…what’s an aspiring inventor in those fields to do? How may such systems be protected? Is all hope lost?

Signed, 
HAL

Dear HAL:

No. All is not lost. We are fortunate that in 1978, the Commission on New Technical Uses (CONTU) helped to modernize United States copyright laws in ways that are still paying benefits to innovators. 

If you’ve been following the Doc’s columns, you already know that copyright protects works, such as books, music, films, photographs, chip masks, software code, and websites. Copyright protects how ideas are expressed (including organization and structure, and how information is captured in software code). Copyright owners get exclusive rights to their works, including the rights to reproduce, distribute, perform, and display. Copyright protection lasts a very long time, as well.

For developers of artificial intelligence and machine learning (“AI/ML”) systems, the law provides protection for many components; the software itself, data used within the system, representations and organizations of information (the “taxonomy”) and others. Copyright on AI/ML source code may be used to prevent others from reproducing the source code, translating the source code into a different computer language, and (in some circumstances) reproducing features of the source code (such as structure of the code), even if not copied verbatim from the protected code. See, Microsoft Corp. v. Buy More, Inc., 703 Fed. Appx. 476, 2017 (where a court granted Microsoft $1,950,000 in statutory damages on a summary judgment); Whelan Assocs. v. Jaslow Dental Lab., Inc., 797 F.2d 1222 (3rd Cir. 1986).

In copyright law, there are no subject matter eligibility requirements as there are in patent law. Where an AI/ML system is ineligible for patent protection, copyright protection can be a powerful alternative tool.

It is also helpful to note that in machine learning, the training data set may be separately protected by copyright as a compilation of data. This means that others may not use the same information to train a system, which protects the machine learning outcomes. 

For an AI/ML system to be legally protected, copyright registrations should always be pursued prior to any publication or disclosure. Copyright is less complex than patenting. Copyrights last far longer than patents. Copyrights are simpler to litigate than patents. Just bear in mind that copyright never protects the idea, but only the particular way that the idea is expressed in a “tangible medium of expression.” If you need to protect an AI/ML concept, a patent is essential to having complete protection.

Are you developing artificial intelligence? Consult the attorneys at LW&H. They are genuinely intelligent and will be glad to help protect your systems.

Until next month,

The “Doc”

— Lawrence A. Husick, Esq.