Developing AI that works for independent producers
AI technologies have been in existence for many years; however, recent advances in computing power and innovative algorithms have resulted in groundbreaking progress, particularly in generative AI. These developments are profoundly transforming the audiovisual industry, offering a range of significant opportunities for innovation in production, postproduction, and audience engagement. At the same time, these developments also present serious challenges that demand thoughtful responses to ensure a sustainable and ethical future for the sector. Among these, the scraping of personal and proprietary data and the indiscriminate use of audiovisual works for AI training without the necessary consent and adequate compensation pose significant risks, particularly for independent producers.
In this paper, CEPI identifies a set of essential elements to ensure that the development of artificial intelligence in Europe is fair to independent producers too.
The importance of copyright compliance
Copyright and related rights represent fundamental legal principles that must be respected by developers and deployers of AI systems. This applies both to the data sets used to train AI systems and to the content generated by them.
Transparency has been identified as a pivotal element in ensuring the accountability of AI systems with regard to rightsholders. By facilitating the identification of the utilization of their work by AI systems, transparency enables audiovisual producers to enforce copyright regulations more easily. For this transparency tool to be effective, AI developers should disclose the content used for training their systems, to ensure that they are not exploiting the IP of rightsholders without consent, or where rights have been reserved.
In addition, rightsholders must have access to an effective mechanism through which they can challenge the unauthorized use of their works in AI systems. The establishment of recourse mechanisms is considered a prerequisite for the effective enforcement of copyright protections and to guarantee appropriate compensation for the use of creative works.
Finally, to protect the integrity of human-created works, content generated by AI is to be clearly labelled as such.
The right to opt-out
It is crucial for producers to maintain complete oversight regarding the utilization of their works within the framework of the text-and-data mining (TDM) exception and AI training. Their right to opt-out of having content included in AI training datasets is a fundamental right that must be respected. This ensures that participation in AI development remains a choice rather than an obligation.
The development of mechanisms to enable independent producers to express their opt-out/right of reserve should be encouraged, as this would allow for greater autonomy and decision-making capacity within the industry. While producers are obliged to make their opt-out declarations clear and machine-readable, it is imperative that the onus is on AI providers to guarantee that reserved works are not used and to verify compliance. This delineation of responsibilities is indispensable for safeguarding the rights of producers while ensuring that AI developers remain accountable. It is the responsibility of AI providers to implement robust systems to verify that the content they use is lawful and in compliance with opt-out declarations or licensing agreements.
Licensing their content
Independent producers who have reserved their rights may yet be interested in allowing their content to be used in AI models. Such use must be explicit and governed by licensing agreements to establish the terms of that use.
The licensing agreements which are utilized in such contexts may assume a variety of formats, and it is incumbent upon producers to select the models that align most effectively with their particular circumstances. This may take the form of a directly negotiated agreement between the independent producers and the AI systems, or a collectively negotiated agreement. Collective licensing can be an option for independent producers, especially for Very Small, Small and Medium Enterprises (VSEs/SMEs), where more leverage becomes possible in negotiating with AI system owners. However, the choice to participate in a collective licensing agreement must invariably remain at the discretion of the producers.
Fair compensation
It is essential that producers who license their works to AI systems receive fair compensation, reflecting the value of their contributions and the value of the work used by AI systems for training purposes, regardless of the ultimate purpose of the AI system.
Empowering producers through AI training
In order to capitalize on the opportunities presented by AI, it is essential that producers possess the knowledge and skills to navigate this evolving technological landscape. Training should focus on the practical application of AI tools to enhance production, streamline workflows, and engage audiences. In addition, it is essential to address the legal and ethical implications of AI use, such as copyright compliance and licensing agreements.
Comentários