Artists leading the fightback against the proliferation of AI

In recent months, voices from all corners of the creative sectors, most prominently within Hollywood, have garnered attention for expressing concerns about the impact of AI on the livelihoods of those at all levels of the industry. Tom Hanks – who had previously been vocal in warning against the dangers posed to artists by AI – was notably the target in a recent unauthorised advertising campaign for a dental plan. Now the future use of AI as a creative tool in literature, art, film, music and photography has been dealt a further blow by two recent developments in the US legal system.

The first of these developments revolves around a growing number of lawsuits filed by writers against AI platform providers, specifically against OpenAI in relation to ChatGPT. The legal reasons for their concerns were detailed in our previous article, but these can be summarised as follows: in order to learn and generate outputs, text-based generative AI applications draw from a range of different sources including books and articles which are “scraped” from the internet. Problematically, these works are often copyright-protected and are used without the permission of their original creators.

ChatGPT has already been the subject of a number of lawsuits in California due to the way in which it generates content. However, OpenAI has argued in its defence that such claims not only overstate the scope of protection offered by copyright to authors, but they also fail to adequately account for exceptions to copyright, most notably fair use, which in the past has proved pivotal in enabling nascent technological innovations to navigate and overcome legal obstacles.

The latest group of concerned authors consists of such prominent literary figures as George RR Martin, Jodi Picoult and John Grisham, who have collectively sold hundreds of millions of copies of their works over the years. This group filed their action in New York, alleging that OpenAI, through its ChatGPT application, has committed “flagrant and harmful infringements of plaintiffs’ registered copyrights”   constituting “systematic theft on a mass scale”. They argue that allowing such applications to continue in this manner risks destroying the US’s literary culture and rendering redundant the crafts of the country’s many talented authors. Authors therefore require a degree of control over the way in which their works are utilised by applications such as ChatGPT.

Compounding this legal headache for generative AI developers is a recent decision from the US copyright office review board, which ruled that an AI-generated image was ineligible for copyright protection as its creation did not involve a sufficient degree of human input. The work in question earned first place at last year’s Colorado state fair competition. However, given the extent to which the AI-based application Midjourney was responsible for the work’s “traditional elements of authorship”, the board regarded the work as lacking human authorship and therefore ineligible for protection. This was in spite of the considerable degree of creative input that the artist, Jason Allen, claimed to have put into the work, which included decisions relating to prompts for the AI system, scene adjustments, and its overall tone.

This ruling is the latest in a line of US cases which have found AI generated works to be ineligible for copyright protection. In February 2023, the US Copyright Office (USCO) similarly denied protection for images featured in an AI-generated comic book named “Zarya of the Dawn” which, like Allen’s artwork, were developed using Midjourney. In making its decision, the USCO pointed out that unlike other tools utilised by artists, the AI platform’s unpredictability prevent users from creating an image which “amounts to the artist’s own original conception”.

Collectively, the cases examined above extend the long-established originality doctrine to the matter of AI-generated creations, revealing a clear unwillingness on behalf of US lawmakers to confer authorship status upon non-human creators. A similar stance is likely to be taken by their counterparts in the UK and Europe, with there seemingly being little room for AI creations in these jurisdictions’ originality standards of “skill, labour and judgement” and “an author’s own intellectual creation” respectively.