Copyright for Literate Robots

Innovation and Economic Growth, Intellectual Property and Copyright and Trademark

Article Snapshot


James Grimmelmann


Iowa Law Review, Vol. 101, pp. 657-681, 2016


Increasingly, written works are read or written by robots. Many copyright cases conclude that robots can “read” works without infringing copyright law. These cases were correct, but discourage online service providers (OSPs) from using humans to analyze online content.

Policy Relevance

Copyright law disallows humans from performing some activities, but allows these same activities on a vast scale when the actors are robots. 

Main Points

  • Originally, authors made copies of works like books, while readers did not make copies; as technology advanced, starting with the VCR, it became common for readers to make copies, but copies made in the ordinary course of reading are “fair use” and do not infringe copyright.
  • Copies do not infringe if the copies are used for another purpose than the original, to add insight or other value; this is called “transformative fair use.”
  • Digital technologies change the quality and quantity of copying; copies are now made for purposes other than expressive reading, and such copies can be made on a vast scale.
  • Programmers can copy a video game to “reverse engineer” it to understand how to make other games that would play on the same console, because the copy is made to learn something about the work rather than to enjoy the work’s expressive content.
  • Making exact copies of a whole work is allowed if the copies are used in a process that does not involve understanding the work’s expressive content, even when many copies are made.
    • Search engines like Google show “thumbnail” images to viewers, but the images do not infringe because they act as pointers for search users.
    • Google Books contains many copies of entire books, but does not infringe because it is not a tool that people use to read books.
  • If a defendant’s copying and analysis is portrayed as a substitute for something that human readers do, the court is more likely to find that the copies are not “fair use.”
  • Computer programs are copyrightable even though they are “read” by computers; but the courts allow broad “fair use,” and Congress created exemptions from liability to let consumers run software and make backup copies.
  • The law governing OSPs’ liability when users post material that infringes copyright depends on whether the OSP had actual knowledge of the infringement, encouraging OSPs to use automated processes to avoid learning of infringement.
  • By exempting reading by robots from liability, copyright law denigrates human reading; it discourages humans from engaging with the expressive content of a work.

Get The Article

Find the full article online

Search for Full Article