With the Canadian federal election coming to its conclusion, many eyes are turning to the last government’s proposals for internet and copyright policy reform. Unfortunately, some of these were quite concerning, such as an “online harms” proposal that the Electronic Frontier Foundation called “dangerously misguided“; all Canadians should learn more about this proposal and use Open Media’s tool to let the government know what you think. Fortunately, other policy actions, such as the Consultation on Artificial Intelligence and the Internet of Things, present more promise.
Internet Archive Canada recently made a submission in response to this AI consultation. As mentioned in the submission—and as recognized in the consultation paper itself—AI raises a fundamental and recurring copyright question: how to ensure the law keeps pace with technological change. In our view, what history suggests as an answer is not interminable legislative tweaks, but rather flexible copyright frameworks, including flexible limitations and exceptions like the fair use doctrine. The Supreme Court of Canada has shown a wonderful and enduring commitment to a flexible conception of fair dealing—most recently in the York University v. Access Copyright case. Why not continue down this path and simply reaffirm the flexible and open nature of fair dealing in Canada today?
The artificial intelligence consultation itself helps prove the point. The Government has been considering taking action on artificial intelligence since a review of the Copyright Act commenced all the way back in 2017 (itself set in motion years earlier). In the years between then and now, should Canada’s technology industry have taken a wait-and-see approach, while others made extraordinary investments in AI? Why spend half a decade or more tweaking narrow legislation when broadly flexible limitations and exceptions can and do fill this gap? And flexibility provides a host of other benefits, including for AI itself. For example, AI is not immune from the ancient maxim, garbage in, garbage out, to say nothing of the bias and other similar problems with AI. As a result, it is important that AI researchers and others be able to analyze datasets both before and after ingestion, and that copyright not stand as an undue obstacle to this work. Legal frameworks empowered by flexible copyright limitations and exceptions, such as controlled digital lending, can help facilitate this process.
In the end, the AI consultation offers much to be thankful for, including an open and transparent process and many good ideas. We look forward to continuing to work with our Canadian friends and neighbors to ensure good copyright policy and strong libraries in the 21st century and beyond.