Coalition for the Diversity of Cultural Expressions (CDCE) (00530)

The information on this Web page has been provided by external sources. The Government of Canada is not responsible for the accuracy, reliability or currency of the information supplied by external sources. Users wishing to rely upon this information should consult directly with the source of the information. Content provided by external sources is not subject to official languages and privacy requirements.

Brief of the CDCE in the context of the Consultation on a Modern Copyright Framework for Artificial Intelligence and the Internet of Things

Presentation of the CDCE

The Coalition for the Diversity of Cultural Expressions (CDCE) brings together the main English and French professional organizations in the cultural sector in Canada. It is composed of some 40 organizations that collectively represent the interests of more than 200,000 professionals and 2,000 organizations and businesses in the book, film, television, new media, music, performing arts and visual arts sectors. The CDCE's main objective is to ensure that cultural goods and services are excluded from trade negotiations and that the diversity of cultural expressions is present in the digital environment.

Introduction

The pandemic has considerably accelerated the digital shift in the cultural sector. Already badly affected by the inability to put in place a system that adequately remunerates rights holders, particularly to take account of the growing supply of online content, the sector is now in the process of collapsing. While one in four people working in the sector lost their jobs in 2020Footnote 1, the digital platforms that distribute cultural content to a growing proportion of the Canadian population have reaped record profits. These companies are profiting from their artificial intelligence systems, which not only allow for the personalized curation of huge catalogs, but also for the analysis of a gigantic amount of data that is, for the most part, jealously guardedFootnote 2.

In its February report, the Standing Committee on Finance (FINA) recommended that the government "complete the review of the Copyright Act later this year by making the necessary amendments to the Act to ensure that rights holders are fairly compensated for the use of their works."Footnote 3 Let’s recall that the GDP of culture exceeded $57 billion in 2019, or 2.7% of the Canadian GDP and represented in 2018 more than 655,000 jobs, far ahead of the agriculture, natural resource extraction, oil and gas, utilities and automotiveFootnote 4.

Some of the questions raised in the consultation document pose fundamental and existential issues for the cultural sector, which has already been disrupted by artificial intelligence systems. While it is difficult to predict the scope and time horizon of the transformations that will affect the sector, the CDCE notes that the questions raised in the consultation document seem to mark out a reflection that is even further away from the purposes of the Copyright Act and that would benefit from being further developed.

In the context of this consultation, we would like to raise issues that seem to be absent from the reflection conducted so far in order to identify actions that will allow rights holders to benefit from the development of artificial intelligence. The development of technologies is inevitable. The real challenge is to contribute to the definition of the objectives that should guide this development.

Finally, let's specify that the questions related to the Internet of Things are not relevant for us, at least for the moment.

1 - The purposes of the Copyright Act

In her analysis of the 2012 reform of the Canadian Copyright Act, Ysolde Gendreau recalls one of the purposes of copyright. "A more comprehensive approach should also make it possible to evaluate the remedies regime in conjunction with the remuneration regimes that are provided for in the law: the effectiveness of copyright is not only judged by its constraint; it is also judged by its institutional capacity to remunerate right holders."Footnote 5 This more comprehensive approach would, in her view, "allow for a more realistic view of the overall effect of the interactions within the same legal concept as part of copyright, be it exceptions, remedies, rights, duration, etc.". She concludes her article by suggesting that "Canadian copyright law is no longer dominated by the interests of copyright owners, but rather by the interests of those who work in the technical world of users of copyrighted works” Footnote 6.

This is exactly the impression we got when we read the objectives of this consultation which, like the two others that preceded it, seems, at best, to give the interests of creators and authors the same importance as those of the burgeoning artificial intelligence industry. Is the cultural sector invited to contribute to the good functioning of another sector of activity, even if it means a decrease in the income of authors and creators and possible investments in new creative projects supported by rights holders? This would be reminiscent of the dynamics at work in the education sector outside Quebec.

We understand, of course, that the government wishes to "support innovation and investment in AI and other digital and emerging technologies across all sectors in Canada"Footnote 7. The Canadian government has, rightly, an arsenal of measures to support this sector. However, by hijacking the objectives of Canada's copyright regime, the latter is in danger of simply becoming ineffective.

2 - Some applications of AI in the cultural sector

The CDCE has been looking at some of the interactions between artificial intelligence and culture for a few years nowFootnote 8. Several other publications, discussing opportunities, impacts or ethical proposals, are worth consultingFootnote 9. In this section, we would like to discuss three examples of technologies that rely heavily on text and data mining (TDM) and contribute particularly to the overall reflection on the issues raised in the consultation paper, in addition to better situating our demands.

First, the recommendation algorithms used by online platforms for streaming or offering cultural content play a curation role that has direct (pay-per-play) and indirect (benefits of listening: live performances, merchandise, purchases of other books by the author, etc.) impacts on the remuneration of rights holders, but also on the diversity of content offered to usersFootnote 10.

However, algorithms are also used to analyze the data generated by users' interactions with the works. According to a group of researchers, "The business of music has, then, morphed to become geared as much around analyzing listening data and crafting discovery algorithms as it is around finding and nurturing emerging and established musicians."Footnote 11 Netflix is also well known for conducting a lot of analytics from user-generated dataFootnote 12. Massive data analytics are used to sell ads, but they are also used to guide the development of cultural content (format, genre, etc.), sign new talents and create new production tools.

These analyses create new added value from cultural works. Thus, the companies that exploit them derive value from 1) listening to or selling works and 2) analyzing the data generated by interactions involving the work in a system.

Second, AI-assisted creation is developing in many areas. Let's take the example of the music industry. Applications such as AIVAFootnote 13 allow the user to obtain entirely computer-generated tracks based on the selection of a music genre or reference to an existing piece whose "emotional impact" one wishes to recreate.  In the visual arts, The Next RembrandtFootnote 14, an initiative supported by ING and Microsoft, generated a novel painting using software and algorithms that analyzed all of the artist's works to extract the most common characteristics from the techniques used by the painter. In journalism, FTD is used to extract facts to generate new storiesFootnote 15.

Third, applications rely on AI and various techniques, including TDM, to generate synthetic mediaFootnote 16: deepfakes, holograms, digital replicas, stand-ins, virtual characters and environments, etc. Applications reproduce the voice and/or image of deceased artists and can make them appear in movies (Carrie Fisher), on stage (Roy Orbison), or read books. Text-to-speech systems can reproduce a person's voice print, while other systems can rework the facial expressions of actors to adapt them to dubbing.

New "synthetic" stars are appearing on social media and becoming models for prized brands: Lil Miquela has three million followers on Instagram where she presents herself as a 19-year-old robot living in LAFootnote 17. She poses for Prada and Calvin Klein, is a musician and "signs" lucrative contracts with companies like SamsungFootnote 18.

Several debates are taking place in the United States, such as to update lawsFootnote 19 to better protect artists from these manipulations when their image or voice is used, and it is often laws other than those on copyright that are invoked for these purposes. But these developments, frequently based on TDM, also suggest a growth in the production of works in which performers are generated by synthetic media.

3 - Exceptions for text and data mining

As we mentioned in the first section, the numerous exceptions to the law lead to a structural imbalance and already deprive right holders of considerable revenuesFootnote 20. The addition of more than thirty exceptions in the last few years, some of which are in contravention of international lawFootnote 21, has completely unbalanced the cultural ecosystems. It would be absolutely deplorable if the next revision were to worsen the situation even further.

We have also seen, in the second section, that AI systems have significant disruptive effects for the cultural sector. In the context of TDM, copyrighted works form a kind of raw material in a value chain from which an innovative product emerges, which is then commercialized. Examples are becoming more and more numerous across many creative sectors. While some rights holders may benefit from these systems, many will certainly lose opportunities to the very AI systems that feed on their creations. The absence of remuneration for the rights holders whose works have contributed to the formation of these systems is particularly unfair.

We therefore call on the government neither to expand the scope of the current exceptions, nor to agree to new ones.

Collecting societies that are members of the CDCE have never been directly solicited for TDM licenses. It is possible that no TDM affecting their repertoire has been done in Canada, that rightsholders have negotiated directly, for example in the literature and newspaper sectorsFootnote 22, or that it has been done without the knowledge of rightsholders.

The CDCE is not in a position to support the implementation of a particular collective licensing regime that would have the effect (or purpose) of substituting for the free negotiation of licenses between rights holders and users.

However, we wish to remind that, since the option of a special regime is raised, the government should imperatively amend the Act to confirm the binding nature of the tariffs set by the Copyright BoardFootnote 23.

Furthermore, copyright owners should be able to accept or refuse that their works be subject to TDM in any eventuality. 

Recommendation 1

That the government neither amend the current exceptions to include TDM, nor implement new exceptions for TDM.

Recommendation 2

That the government amend the Act to confirm the binding nature of tariffs set by the Copyright Board.

Recommendation 3

That copyright owners be given the ability to opt out of TDM for their works.

4 - Ownership of AI-generated works

The issues discussed in relation to the ownership of copyright in AI-generated works raise fundamental questions for the cultural sectors. From the outset, it seems to us that these questions should be addressed in the context of political rather than technical consultations. Moreover, debates on this very issue are ongoing at WIPO and in many other jurisdictions, particularly in the European Union. We believe that Canada would benefit from a better appreciation of the evolution of these debates before considering changes to its law regarding the ownership of works.

For the time being, Canadian law has already demonstrated its ability to deal with these issues by reiterating that only works created through original contributions attributable to human beings can be protected by copyright, by identifying such contributions and their authors at the various stages leading to the creation of works in the context of an AI-assisted creation process, and by attributing ownership of the copyrights resulting from these contributionsFootnote 24.

We therefore consider that the current provisions of the Copyright Act already allow to determine whether or not what results from an AI-assisted creative process or what is generated by an AI is an original copyrighted work and, if so, to identify its author(s) and, therefore, any copyright owner.

In addition, we believe that, at this time, copyright royalties are not an appropriate mechanism to monetize the business model of companies developing artificial intelligence systems for cultural creation. We note that several applications are developed on a subscription basis, although they sometimes offer free products, but with limited functionality. On the other hand, we also note that the market is in the process of providing its own response to the questions raised in this consultation. For example, the company AIVA has decided that the intellectual property of the tracks generated by its application will be modulated according to the subscription planFootnote 25.

Granting copyright to AI systems for autonomous creations, without any original contribution attributable to a human author, would shift the copyright regime from a paradigm of protecting and promoting human creativity, to the pursuit of innovation and revenue for companies of all kinds. This would have far-reaching consequences, the long-term impact of which would be difficult to appreciateFootnote 26. How could the regime reconcile the interests of creators and rights holders with those of very large companies that, thanks to a vertical integration already underway, would be both issuer and purchaser of licenses?

We are also concerned that this would lead to even greater investment in the development of these applications, which would undoubtedly be able to generate an immense amount of work. This prospect of mass production of pseudo-cultural content entirely generated by AI systems should be a major social concern, not to mention the loss of jobs in the sector. Since creation would thus be the result of companies seeking to make their products profitable and no longer of creators, artists and passionate people, bearers of a vision of the world, representative of a great diversity, does cultural content become a product like any other, intended to be consumed by the greatest number? Will the works created by humans be confined to niche markets? How would the significant decrease of creators and artists affect individual and collective capacities for creation and innovation? How could humans learn to be artists or creators without learning from the practices of other artists or creators?

Despite the limitations of creative systems to date, this scenario of mass cultural production by AI is not completely far-fetched. The movement is well underway, and innovations are expected to multiply in the coming years. In May 2018, researchers revealed the results of a surveyFootnote 27 of over 350 AI researchers. On average, they predicted that AI will be able to surpass humans in producing grade-level essays by 2026, popular songs by 2028, and Best Sellers by 2049.

Products resulting from AI generation processes that lack discernible original human input are therefore not "works" protected by copyright and should not become so at the risk of distorting the foundations and objectives of copyright.

Nevertheless, if the legislator were to consider, in order to support Canadian AI research, granting protection to the products of AI generation processes, this protection would have to be based on principles and notions other than those specific to the protection of authors, such as the granting of exclusive or remuneration rights similar to those granted to the producers of sound recordings and, in many countries, audiovisual recordings, while reflecting on who should benefit from these rights. For example, granting the ownership of AI products to the owners of the rights on AI software could have the consequence of concentrating in the hands of a few, probably foreign and quasi-monopolistic, companies a considerable part of the AI products likely to substitute, or even supplant, human cultural creations. A danger that could possibly be partly mitigated by granting instead such protection to the users of these AIs, potentially more numerous and diversified.

We also consider that the services of performers involved in the performance of AI productions that can be assimilated to works should be protected, regardless of the lack of copyright protection for these AI products.

Indeed, the definition of "performer’s performance" found in the Copyright Act specifically referring to the performance, recitation, reading or improvisation of a "work", we recommend that in the event that the government concludes that the production of AI is not a "work" within the meaning of the Copyright Act, the definition of "performer’s performance" in the Copyright Act be adjusted accordingly so as to ensure the protection of performer’s performances of these productions that performers may be called upon to perform.

Recommendation 4

That the government does not extend copyright protection to productions generated by AI without original human contributions even where those productions are comparable to original works.

Recommendation 5

If the government is considering granting protection to the productions generated by AI without original human contributions, something we don’t recommend, then do so through rights similar to those granted to producers of sound recordings.

Recommendation 6

That the performances of performers involved in the performance of productions generated by AI that can be assimilated to works be protected without regard to the absence of copyright protection of these AI productions and that, in the eventuality where the government concludes that productions of AI are not "works" within the meaning of the Copyright Act, the definition of " performer’s performances" which is found in the Copyright Act be adjusted accordingly so as to ensure the protection of performer’s performances of these productions that performers may be called upon to perform.

5 - Promote access to data and data sets

Rightsholders are demanding more transparency from all major platforms in order to be able to provide the necessary information to distribute fair remuneration to their members. While some of the major platforms provide information to rightsholders, not all online platforms provide the same level of information as others. From some online platforms, rightsholders encounter refusal to share datathey consider confidential. The law should be changed to require users to provide rightsholders with the information they need to negotiate more fairly with them.

Recommendation 7

That the Act be amended to require users to provide rightsholders with the necessary information so that they can negotiate more fairly with them.

6 - The revision of the Copyright Act must include other issues than those that will be addressed in the technical consultations

The CDCE has made other recommendations to improve the Copyright Act, which are reproduced in Appendix 2. We estimate the impact of our recommendations to amend the Act at a minimum of $136 million in autonomous revenuesFootnote 28, paid by businesses for the use of content, which could be reinvested each year in the cultural ecosystems that, it must be remembered, generate what seems to be necessary to feed the insatiable appetite of AI, as well as in the Canadian economy. And this does not even include the implementation of certain measures such as copyright term extension, the benefits of which are certain, but which could not be evaluated.

We insist that the next reform also include the other CDCE recommendations, even if they are not subject to technical consultations.

Recommendation 8

Include in the next reform the other recommendations submitted by the CDCE, as reproduced in Appendix 2, even if they are not subject to technical consultations.

Appendix 1: List of Recommendations

Recommendation 1

That the government neither amend the current exceptions to include TDM, nor implement new exceptions for TDM.

Recommendation 2

That the government amend the Act to confirm the binding nature of tariffs set by the Copyright Board.

Recommendation 3

That copyright owners be given the ability to opt out of TDM for their works.

Recommendation 4

That the government does not extend copyright protection to productions generated by AI without original human contributions even where those productions are comparable to original works.

Recommendation 5

If the government is considering granting protection to productions generated by AI without original human contributions, something we don’t recommend, then do so through rights similar to those granted to producers of sound recordings.

Recommendation 6

That the performances of performers involved in the performance of productions generated by AI that can be assimilated to works be protected without regard to the absence of copyright protection of these AI productions and that, in the eventuality where the government concludes that productions of AI are not "works" within the meaning of the Copyright Act, the definition of " performer’s performances" which is found in the Copyright Act be adjusted accordingly so as to ensure the protection of performer’s performances of these productions that performers may be called upon to perform.

Recommendation 7

That the Act be amended to require users to provide rightsholders with the necessary information so that they can negotiate more fairly with them.

Recommendation 8

Include in the next reform the other recommendations submitted by the CDCE, as reproduced in Appendix 2, even if they are not subject to technical consultations.

Appendix 2: List of recommendations made by the CDCE in September 2020Footnote 29

Organizations that benefit from culture need to deliver a greater share of the value generated by copyrighted content 

Recommendations 1 and 2: See CDCE's submission on a modern copyright framework for online intermediariesFootnote 30

Recommendation 3: Abolish the public performance royalty exemption for performers and producers for commercial radio stations.

Restore the balance so that Canadian law allows creators and rights holders to recover royalties on their works

Recommendation 4: Proceed with the immediate implementation, without exception or condition, of the extension of the term of copyright from 50 to 70 years after the death of the author.Footnote 31

Recommendation 5: Amend the fair dealing provisions in the context of education so that they only apply where a work is not commercially available under a licence by the rightsholder or a collective society.

Recommendation 6: improve the private copying regime by

  • Clarifying the Act to make the private copying regime truly technology-neutral;
  • Allowing the payment of royalties for rights holders in the audiovisual, literary and visual arts sectors
  • Eliminating the exceptions introduced in 2012 to sections 29.22 and 29.23 for private copying and time-shifted listening and viewing.

Recommendation 7: Amend the definition of sound recording to include sound recordings that accompany audiovisual works.

Recommendation 8: That Canada ratify the Beijing Treaty and grant rights to performing artists on audiovisual medias in the Act.

Recommendation 9: That resale right be incorporated into the Copyright Act.

Recommendation 10: Introduce a neighbouring right for newspaper publishers.

Strengthening the collective management system

Recommendation 11: Amend the Act to confirm the binding nature of tariffs set by the Copyright Board.

Recommendation 12: Ensure that right holders in the various sectors have the same tools by ensuring that all collecting societies can claim statutory damages of three to ten times the value of the tariff that has not been paid.

Recommendation 13: Raise the upper and lower limits of statutory damages for non-commercial violations and allow the establishment of higher damages in case of systematic and massive use.

Take into account the needs and realities of Indigenous artists, creators and organizations