This highly popular and long running session (presented to a standing room only audience) on the final day of the conference was the 14th time it occurred. It was organized and moderated by Ann Okerson, Director, Offline Internet Consortium. She has moderated all 14 of the sessions.

Ann Okerson

The format was the same as it has been: attorneys from two law firms discussed major legal issues currently affecting the information industry.

Kyle Courtney, Copyright Advisor, Harvard University asked, “Does training an AI system infringe on the works or data used to train the AI?” and “What is the library’s role in this space?” A related question is “What is copyright, really?” Is it just “a bundle of rights with a limited economic monopoly for authors”? The Constitution of the U.S. says that its purpose is “to promote the progress of science and the useful arts”.Here are the rights of the copyright holder.

It is important to note that although copyright holders get some rights, they do not get total protection from infringement; users also get some limited rights. There are some statutory exceptions to the copyright law, and Fair Use is one of them. Fair Use depends on

• The purpose and character of the use,
• The nature of the copyrighted work,
• The amount of the work used, and
• The effect on the potential market value of the work.

According to the Supreme Court, a work is transformative if it adds something new and alters it with a new expression, meaning, or message. Without transformative fair use, we would not be able to quote from an article for comment, view thumbnails of images on the internet, mount online exhibitions from libraries, and conduct several other activities. Licensing offers some control, which is why it is promoted as a solution to AI training resistance. Libraries have large collections, can be transparent about their legally acquired copies, and can help to train reliable and ethical AI.

Michelle Wu from the Georgetown University Law Center focused on contracts, equitable claims and AI using as examples of contracts involving code, legal opinions, and photos plus metadata. She listed safest uses for libraries and risky ones;

AI has opportunities and misuses; we must anticipate both and prepare for both. 

Don Hawkins, Conference Blogger