Copyright and innovation by Matthew Marinett
Jul 5, 2017
If Canada is to become an major centre of high-tech business and AI development, it must remove the copyright-related impediments to innovation.
Source: Policy Options
fter a lengthy decline in its standing, Canada is attempting to reassert itself as a major centre of high-tech businesses and innovation. Bolstered by concerted efforts by the federal government as well as by regressive immigration policy south of the border, Canada is positioning itself to become a haven for technology start-ups and talent.
However, for Canadian tech centres to be competitive with international innovation hubs like Silicon Valley, we must ensure that our laws reward innovators without closing the door to new start-ups and disruptive innovation. This has always been the central trade-off of intellectual property law: by providing exclusive rights to a creator, we may ensure just rewards for her creation, but we may also prevent others from using and building upon that creation. Finding the proper balance is key to promoting sustained innovation, but it’s no easy task, as the balance is constantly disrupted by new technologies.
This is why regular review of the Copyright Act is so important, and why the upcoming mandatory review, and each subsequent one, should be seen as an opportunity to continue to promote innovation rather than merely debate past changes. Indeed, as Michael Geist has already written in this Special Feature, the 2012 reforms were largely successful in bringing Canadian copyright laws into compliance with international obligations and up to speed with existing digital technologies like the Internet. But while those changes were laudable, the existing Copyright Act leaves unaddressed or unclear a number of pressing issues for the future of Canada’s innovation and technology sectors. Here are three that must be dealt with in the upcoming review.
What rights do purchasers have in goods with embedded software?
Normally, consumers are free to modify or repair the physical goods they own. However, as copyright-protected computer programs are being incorporated into more and more everyday objects, from cars to consumer appliances to smart houses, this freedom may be threatened by the current incarnation of the Copyright Act.
The danger arises from the copyright protection of software combined with provisions added in the 2012 copyright reforms that prohibit the circumvention of technological protection measures (TPMs). The Copyright Act defines a TPM, commonly known as a digital lock or digital-rights management, as “any effective technology, device or component” that ordinarily prevents access to, or the copying of, a copyright-protected work. The Act currently makes it unlawful (with a few exceptions) to circumvent TPMs to gain access to a protected work, even if that access would otherwise be lawful. In the world of ubiquitous embedded software and the rapidly growing Internet of Things, this prohibition could become a powerful tool to limit the rights of consumers and prevent innovation and competition.
A smart home, for example, may be equipped with TPM-protected software that controls the climate, lighting and power consumption of each room. If problems arise that require accessing the software to repair, the homeowner might be required to hire a repair person authorized by the copyright owner, and would be prevented from making changes herself. The same requirement might apply if the homeowner wishes to make any modifications to the system or add additional elements. In each case, when the software needs to be accessed, the homeowner appears, under the current legislation, to be tied to the copyright owner.
In the US, the issue has already arisen in the context of automotive modifications and repair. Software built into vehicles has now been exempted from the anti-circumvention provisions in the Digital Millennium Copyright Act, but there are no general provisions to exempt all goods with embedded software.
These restrictions clearly have enormous ramifications for aftermarket businesses and the repair industry, but they are also fatal to a great deal of innovation, for at least two reasons. First, and most obviously, innovation often entails tinkering with existing products to improve them or modify them to perform new functions. As physical objects increasingly become embedded with software, their manufacturers might try to prevent such tinkering by including TPM-protected software that shuts off any avenue of modification.
Second, and most frighteningly, software can be used to tie one consumer product to other products manufactured by the same company. Imagine that the software in your smart home rejects any light bulb that isn’t manufactured by the software provider’s partners. While the Copyright Act allows circumvention of TPMs for the purpose of making computer programs interoperable, it does not include an exception for making physical things interoperable. So long as the software that prevents the use of other light bulbs is itself protected by a TPM, modification of the software to allow the use of third-party light bulbs might be unlawful. If companies that design parts for larger systems are excluded from the market by this sort of vertical integration, the impact on innovation would be profound.
It may well be that future courts will interpret the TPM provisions narrowly or recognize exceptions in the law to the prohibition on circumvention. Some experts think the TPM provisions might be unconstitutional. However, in the only case so far that substantively addressed the TPM provisions, the court read them broadly and applied them forcefully, going so far as to hold that circumventing even physical barriers to accessing software violates the TPM provisions.
Given the real danger to the future of Canadian innovation posed by these provisions, it is imperative that we address not only how TPMs are used to protect digital entertainment products but also how they are likely to be used in the software-embedded world of the present and the very near future. The review of the Copyright Act may be an excellent place to begin this process. Specifically, the review should consider implementing new exceptions to the TPM provisions for the modification, repair, improvement and interoperability of goods with embedded software.
What data can artificial intelligence systems use?
A major focus of Canada’s innovation strategy is the development of artificial intelligence (AI) and machine learning. But if Canada is to become a hotbed of AI innovation, it must make sure the legal environment allows for the flourishing of AI systems themselves. One key characteristic of AI and machine learning systems is that they often become more powerful (and useful) the more information they are able to collect and analyze.
Existing copyright law may hobble this development process by preventing these systems from acquiring the information they need to operate effectively. Many potentially useful datasets, including those derived from the Web, are protected by copyright law, which might prevent AI systems from legally scanning and analyzing them. While the 2012 reforms allowed “temporary reproductions for technological processes,” it’s not clear that this provision would apply to text and data mining carried out by AI. Nor is it at all clear that such a use by AI systems would qualify as fair dealing, a defence that does not usually apply when copying is done for commercial purposes.
To avoid the possibility of a chilling effect on AI innovation in Canada, the government must create a clear exception to copyright restrictions that will allow text and data mining by AI and machine learning systems in Canada. It should be included in new reforms following the review of the Copyright Act.
Who owns the works created by artificial intelligence systems?
Among their many applications, one use of AI and machine learning algorithms is to produce artistic works, or even write tailored computer programs. One unanswered question that will become increasingly important as AI is more broadly deployed is whether these creations are subject to copyright, and if so, who owns them.
In one famous American case, a court has already taken a stab at a similar issue. The “monkey selfie” case made international news, largely for its amusing focus on a series of images taken by a macaque. More broadly, the case was also newsworthy for addressing the question of whether copyright subsists in works created by nonhumans. The answer in the US, it turns out, is no.
In Canada, the question remains open. Subsection 5(1) of the Copyright Act establishes “conditions for subsistence of copyright” that strongly suggest that only corporations or natural persons can be the authors of a copyright-protected work. But in the case of widely deployed AI systems, multiple natural persons might claim authorship. Currently Canadian courts would decide who is the author based on which persons exercised sufficient skill and judgment to produce an “original” work. But this test leaves an important issue uncertain: if neither the user nor the creator of the AI system can demonstrate the necessary skill and judgment to be the author of the work produced by the AI system, is the work simply outside the scope of copyright protection?
How we answer that question may have a powerful effect on AI innovation. Allowing works created by machine to fall outside the scope of copyright protection may promote a robust user-based ecosystem of free works, but it may also undermine incentives for AI developers to produce the systems that generate those works, for two reasons. First, AI developers won’t be able to commercialize the outputs of their AI programs. Second, even if an AI firm intends to market only the AI program itself, and not the products it creates, consumers and corporations are less likely to pay significant sums for a system that produces works they cannot commercialize. Although AI-created works are of limited commercial importance at the moment, the question of ownership should be included in the Copyright Act review.
Keeping pace with innovation
Naturally, the Copyright Act can be only one small piece of Canada’s innovation agenda, but it remains an important one that can’t be overlooked. Copyright is capable of both promoting innovation and stifling it. And if Canada seeks to compete in the rapidly advancing high-tech fields, it cannot afford to simply react to developments. If we wait for the next five-year review to remove impediments to innovation in areas like AI and data science, we may well be too late to join the forefront of AI development. To be technologically innovative, Canada must be legally innovative as well, by making forward-looking reforms.
The federal government has provided few details on how the review will be carried out. It has not indicated whether the legislation is being opened up for significant reform. If the review results in nothing more than some cursory consultations without action, it would be a missed opportunity. For Canada to become an innovative nation, it must promote innovation at every occasion.