Screen Composers Guild of Canada (SCGC)

Les informations de ce site Web à été fournie par des sources externes. Le gouvernement du Canada n'assume aucune responsabilité concernant la précision, l'actualité ou la fiabilité des informations fournies par les sources externes. Les utilisateurs qui désirent employer cette information devraient consulter directement la source des informations. Le contenu fournit par les sources externes n'est pas assujetti aux exigences sur les langues officielles et la protection des renseignements personnels.

September 17, 2021

Submission of the Screen Composers Guild of Canada
In response to: Consultation on a Modern Copyright Framework for Artificial Intelligence and the Internet of Things

  1. The Screen Composers Guild of Canada (SCGC) is the national association certified under the Federal Status of the Artist Act to represent all professional Anglophone composers and music producers for audiovisual media productions in Canada.
  2. As outlined below, SCGC respectfully submits that:
    1. Assigning authorship to AI-enabled works depends on who makes the underlying creative decisions and should be guided by current practices for assigning authorship/ownership to works that contain samples or excerpts of existing copyrighted music.
    2. Regardless of how efficiently AI technology can examine and extract portions of countless existing works, the creators of those existing works being used as source inputs are still the authors and primary owners of those works, and must be credited and compensated accordingly by the creator of the new AI-assisted work.
    3. Even where the author of the new AI-enabled work remains the programmer or creator who trained or used the computer to output new work, the owners of the source material being analyzed, interpolated, extrapolated, ‘mashed-up’, sampled or emulated must be treated as co-authors and be compensated for any royalties triggered by the use of their compositions or sound recordings in the AI-enabled creative process. Even if the outputs become the property of the programmer who selected the inputs, the inputs remain the property of the artist(s) who created them.
    4. Therefore, there is no need for new copyright exceptions for use of copyrighted music by AI platforms, for any purpose. If new approaches for requesting and providing consent from rights holders are required to reflect the computing capacity of AI technologies, they should be based on the notion of explicit permissions, rather than automatic permissions. 
    5. AI technology should be regulated in accordance with existing principles of transparency and accountability. Regulatory responses to AI technology should be configured to promote meaningful discovery of Canadian content, and to support proper remuneration by rights holders for use of copyrighted music.
    6. Regulations governing the use of AI technology in a copyright context should ensure that the genealogy of new works (including data describing source materials and cultural content inputs) are captured and disclosed within the meta-data of the new work being created with the assistance of AI technology. This is the only way that rights holders in the underlying works can be sure they are being properly compensated for use of their work.
    7. Failure to disclose ownership over any constituent element of an AI-enabled work, whether knowingly or otherwise, should be treated the same way under law as any other uncredited use of another person’s copyrighted material.
    8. The principle of algorithmic transparency should apply to the use of AI technologies to allow for the discoverability of particular works (such as Canadian-authored works) over others where cultural policies are mandated.

Introduction

  1. All creators and consumers benefit from innovation.  AI technology, however defined, will undoubtedly provide artists, composers, and songwriters with new means of creating art and music, just as it will provide publishers, distributors, and other intermediaries with new ways of identifying and engaging with audiences. However, advances in technology must not be an excuse to erode the natural rights of authors/owners to participate fully in the economic benefits of their work.
  2. Content creators and copyright holders are already faced with determined efforts by media producers, media broadcasters and distributors, and other intermediaries to disenfranchise authors/owners from their inherent legal and ethical rights under copyright policy and law. These efforts range from forcing composers and songwriters to permanently surrender their rights to the work they create as a condition of engagement (i.e. forced buy-outs), to intermediaries claiming to be the ‘creators’ of musical and audio-visual works just because they have commissioned or licenced the underlying intellectual property from the actual authors/owners.
  3. This ongoing review of the copyright framework has underscored the need to enshrine, clearly, in legislation the role of creator as the sole “author” and original owner of any copyrighted work, and to prohibit activities that dehumanize the process of artistic creation.  The fact that term of “ownership” over an original work is based on the life of its author, plus 50/70 years, underscores the irrevocable element of human agency in copyright ethics and law.
  4. Commissioning someone to compose a song or score does not mean that the payee wrote that song or score. Programming a computer to produce a song or score based on the characteristics of a hundred copyright protected songs and scores does not mean that the computer’s programmer is the author of any of those other songs or scores.  It follows that any new works that emerge from an AI-enabled process, based on interpretation and extrapolation of existing works for source material, must duly credit and compensate the originators of the original copyright-protected source materials.

Establishing ‘authorship’ within AI-enabled works should be informed by existing principles for establishing authorship for new works that are based on, or incorporate elements of, existing works

  1. AI technology is increasingly present in the generation and interpretation of cultural content. The role of AI technology in ‘creating’ new works underlines the need to reiterate and reinforce in any updated Canadian copyright framework, the notion that “authorship” requires human agency, and that the “author” of a work is first and foremost its “creator,” who is also the sole and primary “owner” of that work.
  2. Along these lines, SCGC agrees with the Ivors Academy’s highly relevant conclusions on these questions, as submitted to a recent WIPO consultation:Footnote 1
    • The draft issues paper repeatedly speaks of “AI-generated work” and … states that AI applications are capable of producing literary and artistic works autonomously. We urge caution in the use of this terminology, and the assumptions that follow.

      “AI-generated” can refer to a vast range of works in which the ratio of AI use to human contribution can vary drastically. Depending on such proportion, copyright will be applied differently, meaning that the same rule cannot be applied to all ‘AI-generated’ works … As it stands, “AI-generated” musical works always require human contribution.

      Human creation occurs in two phases. First, the AI application needs copyrighted works which are entirely human generated as input material to analyse patterns and machine-learn composition. Second, human input is fundamental in using or ‘training’ the AI application to do this.

      These are crucial considerations both as regards the authorship/ownership of any new copyright works created with the use of AI, and the potential infringement of the copyright in existing works.

  3. SCGC submits that whether ‘generated’ or ‘assisted’ by AI technology, every new work that incorporates or ‘samples’ elements of existing work must credit and compensate the creator/author(s) of that existing work.
  4. This requirement should apply whether the ‘existing work’ is a musical composition or a sound recording.  In either case, any use of copyrighted creations engages any number of existing copyrights, and triggers the payment of any number of royalties, depending on the particular nature of the material and how it is being used. The fact that a copyrighted work is being used to train a computer to emulate it does not obviate the need to credit and compensate the owners of the input material for that use.
  5. Whether AI is being deployed to emulate the characteristics of hundreds of existing compositions in a new composition, or to assemble snippets of hundreds of existing sound recordings into a new sound recording, in either case the authors of those existing works must consent to their use, and if they agree, be credited as co-author of the new work.
  6. This fundamental element of copyright policy is as true today as it was in an analog world. If a composer took sections of other composers’ scores and assembled them into a new score by hand, the original composers had to give their consent and be credited before the new composition could be commercialized.  If a musician took a razor blade and scotch tape and spliced together samples of existing taped recordings, the creators/rights holders of those existing sound recordings had to give their consent before the new recording could be commercialized.
  7. The fact that AI technology is capable of emulating and splicing vastly more existing compositions or recordings, does nothing to change the fact that the person inputting that material to an AI platform is using the copyrighted work of other creators as a constituent element of their own work, and therefore must ask permission, credit and compensate the rights holders of that source material accordingly.
  8. In this regard, SCGC notes the recent interview in Wired magazine, with Grammy-nominated composer BT:Footnote 2
    • I split artificial intelligence into two categories, generative and assisting… Imagine we turn an artist I love, Prince. into a training set. His intellectual property rights are owned, and his estate is fairly remunerated, right? All those i’s are dotted, and the t’s are crossed.  And from that training set the AI could generate new music.
  9. The fact that AI technology can perform that analysis of the source material faster and with a greater degree of granularity than previous computers does not erode the fundamental rights of the owner of the source material.  The owners of the source material being analyzed, interpolated, extrapolated, mashed-up, sampled or emulated must still give their consent and be compensated for any royalties triggered by the use of that music in the AI-enabled creative process. Even if the outputs become the property of the programmer who selected the inputs, the inputs remain the property of the artist who created them and/or their legal rightsholders.

Therefore, there is no need for new blanket exceptions for use of copyrighted material by AI technology.

  1. SCGC submits that there is no credible case to be made for the establishment of broad exceptions to permit unremunerated use of copyrighted material by AI technology, or the people who create/program it.
  2. Recognizing that the sorts of AI-enabled processes contemplated in this proceeding represent an exponential increase in computing capacity, SCGC recognizes that a case can be made for the establishment of an alternative method for obtaining consent and calculating compensation for mass use of copyrighted materials by AI technology or its programmers.  Those approaches should nevertheless be based on the notion of explicit permissions, rather than automatic permissions, just as all other uses of copyrighted work are, when they are used in the creation of a new work.

AI technology should be regulated in accordance with existing principles of transparency and accountability

  1. Any new tool that facilitates and inspires new creative works, and/or new ways of creating new works, clearly has the potential to make a tremendous contribution to Canada’s cultural policy objectives.  Harnessing that potential relies on an appropriate regulatory framework to ensure the new tool contributes to, and does not detract from, existing cultural policy goals.
  2. In this context, the determination of authorship of an AI-enabled work depends, as submitted above, on who exercises human creative agency over the constituent elements of the work in question. It follows that determining authorship – and enforcing related rights – depends on administrative visibility over the source material and other creative inputs.  Properly configured, regulations governing the use of AI technology in a copyright context will ensure that data describing cultural content inputs are captured and disclosed within the meta-data of the new work being created with the assistance of AI technology.
  3. In other words, for an AI-enabled piece of music to be registerable with a rights-administration organization, the registrant should be required to seek permission for their use of source material and include in the registration data information identifying the creators/owners of any constituent elements of the work, allowing them to be compensated.
  4. Failure to disclose ownership over any constituent element of an AI-enabled work, whether knowingly or otherwise, should be treated the same way under law as any other uncredited use of another person’s copyrighted material – the rights holders should have recourse to dispute resolution via the performing rights organization and/or restitution via the courts.

The same principle applies to use of AI to promote discoverability of existing works

  1. An often-noted benefit of AI technology is its ability to make meaningful and predictive recommendations for users, based on a deep reading of their previous choices and consumption habits. Given the ability of content distributors and other intermediaries to program those underlying algorithms to shape and steer content choices, so-called ‘discoverability’ algorithms by definition both facilitate and limit consumer choice, at both a macro and micro level.
  2. Recent legislative proposals in the privacy arena have adopted the notion that consumers should have the right to know how any digital platform collects, interprets, and uses their data to make marketplace recommendations for that consumer.  AI technology’s potential to shape consumer choice and marketplace behaviour is no less pronounced in a music and audio-visual context than it is in a financial, advertising or retail context.
  3. Accordingly, regulatory approaches to AI in a copyright context should be guided by the same principles of algorithmic transparency that are emerging as cornerstones of privacy policy in the digital age.

Conclusion

  1. To summarize, SCGC respectfully submits that:
    1. Assigning authorship to AI-enabled works depends on who makes the underlying creative decisions and should be guided by current practices for assigning authorship/ownership to works that contain samples or excerpts of existing copyrighted music.
    2. Regardless of how efficiently AI technology can examine and extract portions of countless existing works, the creators of those existing works being used as inputs are still the authors and primary owners of those works and must be credited and compensated accordingly by the owner of the new AI-assisted work.
    3. Even where the author of the new AI-enabled work remains the programmer or creator who trained or used the computer to output new work, the owners of the source material being analyzed, interpolated, extrapolated, ‘mashed-up’, sampled or emulated must be compensated for any royalties triggered by the use of their compositions or sound recordings in the AI-enabled creative process. Even if the outputs become the property of the programmer who selected the inputs, the inputs remain the property of the artist(s) who created them or their legal rightsholders.
    4. Therefore, there is no need for new copyright exceptions for use of copyrighted music by AI platforms, for any purpose. If new approaches for requesting and providing consent from rights holders are required to reflect the computing capacity of AI technologies, they should be based on the notion of explicit permissions, rather than automatic permissions.
    5. AI technology should be regulated in accordance with existing principles of transparency and accountability. Regulatory responses to AI technology should be configured to promote meaningful discovery of Canadian content, and to support proper remuneration by rights holders for use of copyrighted music.
    6. Regulations governing the use of AI technology in a copyright context should ensure that data describing cultural content inputs are captured and disclosed within the meta-data of the new work being created with the assistance of AI technology. This is the only way that rights holders can be sure they are bring properly compensated for use of their work
    7. Failure to disclose ownership over any constituent element of an AI-enabled work, whether knowingly or otherwise, should be treated the same way under law as any other uncredited use of another person’s copyrighted material.
    8. The principle of algorithmic transparency should apply to use of AI technologies to promote discoverability of particular works over others.
  1. SCGC appreciates the opportunity to share its views in this important proceeding.

***End of Document***