Artificial intelligence (‟AI”) technologies, which developed exponentially in the last 5 years, are here to stay and thrive. Most legal frameworks, in particular the French and US ones, are not ready for these technological advances. Indeed, most courts still refuse to grant copyright protection and ownership, to AI-generated works, worldwide. This situation is not sustainable, as AI-generating tools and platforms will replace traditional methods of generating content, in a very short timeframe. Legal frameworks must therefore adapt and yield, in order to ensure that their national creative industries remain competitive and at the top of the class. How can this be achieved?
1. The AI revolution: making AI the new normal
1.1. AI technological advances
AI technologies have exponentially developed, in 2022, making it possible to generate various creative outputs, in the literary field, music sector, art and illustration sectors, the film industry, the graphic novels’ sphere and the video gaming industry.
For example, ChatGPT is a language model developed by OpenAI, a research organisation that specialises in developing cutting-edge AI technologies, founded in 2015. ChatGPT is based on the Generative Pre-trained Transformer (‟GPT”) architecture. Its self-described purpose is to ‟provide conversational assistance and answer a wide range of questions on various topics, from factual information to subjective opinions and advice”. It can be used as a powerful research tool, writing extremely coherent and detailed research reports, which may then be inserted verbatim in any article, research paper or blog.
AI writing tools have become commonplace, and now assist humans in a variety of ways, in order to write better emails, blogs, articles and novels, to curate email newsletter content, to generate ready-to-write SEO outlines, to generate articles that are SEO friendly, etc.
In the graphic design sphere, AI tools can produce outstanding illustrations and drawings just by merely giving written instructions to AI tools such as Dall-E 2 (also created by OpenAI), Midjourney, and Stability.ai. The results are outstanding, as Midjourney’s community showcase demonstrates.
These illustrations and drawings can, in turn, be used to create graphic novels, such as ‟Zarya of the Dawn” by Kristina Kashtanova (generated with Midjourney), video games, magazines’ covers, such as The Economist’s cover and Cosmopolitan’s cover for its AI issue, or films, such as ‟The Crow”, a short film generated with OpenAI’s CLIP, and which won the 2022 Jury Award at the Cannes Short Film Festival!
In the music field, AI-generated music is become more prevalent, with AI music composition tools such as MusicLM, a model generating high-fidelity music from text descriptions from Google Research lab, Riffusion, an AI that composes music by visualizing it, Dance Diffusion, Google’s previous project AudioLM and OpenAI’s Jukebox. Soundful, a human aided AI music platform, aims at fulfilling the demand for music within the creator community and content-creator economy. At the click of a few buttons, you can have the tune you need for your project.
1.2. Who are the main players in the AI field?
As mentioned above, OpenAI, is at the forefront of AI innovation, with multiple AI tools created to be exploited on various creative medium such as:
- the spoken word, via ChatGPT;
- films, with CLIP;
- illustrations, via Dall-E 2, and
- music with Jukebox.
OpenAI was founded in 2015 by a group of prominent tech industry figures, including Elon Musk, Sam Altman, Reid Hoffman, Ilya Sutskever, Peter Thiel, John Schulman, and Wojciech Zaremba. It is an American AI research laboratory consisting of the non-profit OpenAI Incorporated (OpenAI Inc.) and its for-profit subsidiary corporation OpenAI Limited Partnership (OpenAI LP). The goal of OpenAI is to create safe and beneficial AI ‟that can help to address some of the world’s most pressing challenges” (sic). OpenAI is an independent organization, and its research is funded by a mix of philanthropic contributions, corporate partnerships, and government grants. The organization is ‟dedicated to advancing AI technology while also promoting transparency, collaboration, and ethical considerations in the development and deployment of AI”.
In 2023, OpenAI announced a partnership with Microsoft. On 23 January 2023, Microsoft announced a new multi-year, multi-billion dollar (reported to be USD10 billion) investment in OpenAI. Then, on 7 February 2023, Microsoft announced that it is building AI technology based on the same foundation as ChatGPT into its web search engine Bing, its web browser Edge, its productivity software Microsoft 365 and other products.
Google is also prominent in the AI-tools’ manufacturing sector, in particular with its above-mentioned music-generating AI tools, MusicLM and AudioLM. On 6 February 2023, Google announced an AI application similar to ChatGPT (Bard, a conversational AI chatbot powered by Google’s Language Model for Dialogue Applications), after ChatGPT was launched, fearing that ChatGPT could threaten Google’s place as a go-to source for information. Google also launched Imagen, a ‟text-to-image diffusion model with an unprecedented degree of photorealism and a deep level of language understanding” (sic), to compete with Dall-E.
For now, the media’s general impression is that Google fell behind in AI, in particular compared to Microsoft and OpenAI.
2. AI’s legal challenges
2.1. Do AI users have rights to the outputs?
The question of AI authorship is particular important, especially if actors from the content-creator economy embrace it.
The first port of call is to review the terms and conditions of the AI-generating tool, in order to clarify who owns what, as far as the AI-generated output is concerned.
For example, in order to use OpenAI’s platforms, such as Dall-E 2, Jukebox or ChatGPT, the user must first agree to OpenAI’s terms of use. In the version dated 14 March 2023 of these T&Cs, it is set out that OpenAI assigns to the user all its right, title and interest in and to the output generated and returned by the OpenAI services based on the user’s input. This means the user can use the Content (defined as the input and the output together) for any purpose, including commercial purposes such as sale or publication, if the user complies with OpenAI’s terms. OpenAI may use the Content to provide and maintain its services, comply with applicable law, and enforce its policies. The user is responsible for the Content, including for ensuring that it does not violate any applicable law or OpenAI’s terms of use.
This is an improvement on the OpenAI’s terms of use which were in force before 14 March 2023. These terms used to state that users assigned any ownership they had in any output created by the OpenAI’s system and services, and, in turn, the users had an exclusive licence to use the generated output for any purpose. Other AI platforms which can generate output, such as Soundful, still have similar contractual copyright and licensing arrangements.
2.2. Who is the author of the AI-generated output?
Was it the person who input the text prompt? Was it the AI? Was it the developer of the AI or the company that owns the AI?
Most jurisdictions require a human to be the author, and a work is only capable of being protected by copyright if it shows intellectual effort, creativity, and reflects the author’s personality.
As AI creative systems become more widespread, can we consider that a text prompt, such as ‟a cat wearing a turban gazing at the city landscape at night, from a window, in the style of Van Gogh”, constitutes enough human input and individuality, and is sufficiently creative and reflective of the human author’s personality to allow the resulting image to be protected by copyright?
For example, UK law permits copyright protection of computer-generated works, with the author being the person who made the ‟necessary arrangements” for the creation of the work pursuant to section 9(3) of the Copyright, designs and patents act 1988. Other rare jurisdictions expressly provide for copyright in computer-generated works, such as Hong Kong, India, Ireland, New Zealand and South Africa.
However, most countries, such as France, refuse to acknowledge copyright protection if the work is generated by anyone other than a human. Indeed, pursuant to article L. 112-1 of the French intellectual property code, ‟any work of the mind, regardless of its kind, form of expression, merit or purpose”, is eligible for copyright protection. French courts acknowledge the originality of a creation as soon as the said creation is endowed by the personality of their author. While the threshold of the originality requirement is low, the author of a work must be a natural person according to a well-established case-law. It cannot be a legal person, an animal or a software. The rationale behind this position is that French law only protects works of the mind and the creations at issue must bear the imprint of the personality of their author(s); legal persons, animals and AIs neither have a conscience nor have a personality that may come out of the works created by them.
Time and time again, the United States Copyright Office (”USCO”) which is in charge of registering works for copyright protection in the USA, refused to grant copyright protection to AI-generated content, such as:
- a picture autonomously created by a computer algorithm entitled ‟A recent entrance to paradise”, which Steven Thaler, the CEO of Imagination Engines, Inc. requested the USCO to register as part of his application dated 3 November 2018, and
- the above-mentioned graphic novel ‟Zarya of the Dawn”, because the images generated by Midjourney, contained within the work, are ‟not original works of authorship protected by copyright” (sic) and since ‟text prompts” are insufficient to qualify as ‟human authorship”.
Since copyright protection is automatically granted in France, unlike in the USA where copyright protection is granted solely upon the USCO’s registration, no such case law exists in France or other European countries. We will have to wait a few more years, before copyright infringement litigation, relating to AI-generated content, lands in the European courts’ dockets (with the notable exception of the claim filed by stock image supplier Getty Images, against Stability.ai, for copyright infringement, with the High Court in London, UK).
Considering the absence of a work of the mind and the lack of originality of an output resulting exclusively from an AI, such creations are thus today in the public domain and not intellectual property right is attached to them (with the notable exception, above-mentioned, of the regime applicable in the UK, Hong Kong, Ireland, India, New Zealand and South Africa, which permits copyright protection of computer-generated works, with the author being the person who made the ‟necessary arrangements” for the creation of the work).
2.3. How can the law adequately address AI, in order to make it beneficial for the creative industries?
In France, the ‟Conseil Supérieur de la Propriété Littéraire et Artistique”, an independent advisory body advising the French ministry of culture and communication in the field of literary and artistic property, provided recommendations in a report dated 27 january 2020. It suggests that a ‟sui generis” right could be created to the benefit of the one bearing the risks of the investment, like the specific regime benefiting to database producers.
The European Parliament, from the European Union (‟EU”) also suggested that a legal personality be acknowledged to IA, so that copyright protection be granted to AI-generated works.
For the time being, none of the above recommendations have been taken up by the French and/or European legislators.
The European Commission (‟EC”) has proposed a four-step test in order to assess whether AI-generated output can qualify as protected work under the current EU copyright framework, as follows:
- step one: the AI-generated output must be a production in the literary, scientific or artistic domain (article 2(1) of the Berne Convention for the Protection of Literary and Artistic Works);
- step two: the AI-generated output must be the result of human intellectual effort (i.e. some sort of human intervention such as, for example, development of software, editing or gathering or choice of training data);
- step three: the AI-generated output must be original, and
- step four: the work needs to be identifiable with sufficient precision and objectivity (i.e. the generated output will fall within the creator’s general authorial intent).
It is blatant that the current disparities in the various legal systems and frameworks, where one set of national legal frameworks strongly pushes back against granting copyright protection to AI-generated content, while the other set of national legal frameworks fully embraces AI-generated content and grants it full copyright protection, will create substantial inequalities of treatment for content creators worldwide.
Indeed, content creators based in the UK or Ireland or India are incentivised to speed up their work flow, by fully embracing time-saving AI-generator tools and programmes, the output of which these creators will be able to claim copyright ownership and protection on. Meanwhile, creatives based in France or the USA will seriously struggle to enforce any type of copyright protection and rights over AI-generated content. Therefore, French and US content creators will prefer to stick to the ‟old methods” of work creation, refusing to use AI platforms which output will automatically fall into the public domain.
This situation cannot perdure, and since AI is here to stay – big time – stubborn law makers and enforcers, such as the USCO which recently issued some narrow-minded guidance on copyright registrations involving AI, will inevitably have to yield and reform, in order to grant copyright protection to AI-generated works.
Of course, lobbies and organisations representing music and other creative disciplines will try to delay the inevitable, by setting up mindless campaigns and lobbying plots, such as the ‟Human Artistry Campaign”, to prevent copyright offices and courts from finding that AI-generated works are protected by copyright.
But the floodgates are now open. And there is no way back: the technological advances and jumps that AI-generating tools and platforms provide are so important and ground-breaking, that the creative content economy and stakeholders at large will fully embrace them in the next few months, never looking back.
Lawmakers must adapt fast, in order to keep their creative economies and actors at the forefront of competition.
Crefovi regularly updates its social media channels, such as Linkedin, Twitter, Instagram, YouTube and Facebook. Check our latest news there!
