Australian authors have been caught up in a dataset of pirated ebooks, called Books3, which has allegedly been used by Meta to train generative AI.
While we await the legal implications of tech companies using copyrighted, creative works, an RMIT expert warns that tech companies have always operated this way – it’s just that now, the game has changed and the stakes are higher.
Dr Dana Mckay, Senior Lecturer in Innovative Interactive Technologies, School of Computing Technologies
“Tech companies are accustomed to using data found on the internet without considering ownership – such as indexing text for search engines.
“There are good historical reasons for this. Without indemnity for both indexing and linking to content, that might be in itself illegal, search engines wouldn’t exist.
“For the most part this situation has benefitted creators and search engines – links to books are advertising for those books.
“Where search tools have been created specifically to provide access to copyrighted content without payment, those tools have faced successful legal challenges (such as the Napster case over 20 years ago).
“For tech companies, using books to train AI models may seem a natural next step from search indexing: ingest content, do their thing with it, present content to the user.
“For authors, artists and other creatives, though, this breaks the link to their original content, thus eliminating any opportunity to track or receive fair compensation for use of their works.
“Meanwhile, AI companies are monetising the use of those works, up to and including producing work ‘in the style of’ a creator.
“Use of copyright works comes with the responsibility to attribute and pay for that use.
“If we, as a society, want the joy, beauty, and humanity that creative works inspire, we need to make the key distinction between linking to work, and ingesting it.”
Dr Dana McKay studies the intersection of people, technology and information, and her focus is on ensuring advances in information technology benefit society as a whole.
