There is no question—2023 was the year Generative Artificial Intelligence (GenerativeAI) exploded in the popular consciousness. As AI tools for generating visual, audio, and written content become more widely available and user-friendly, it is not impossible that GenerativeAI outputs will wind their way into more of the artistic, literary, dramatic, and musical works we produce both mundane (this article?) and sublime (award winning art). It's already happening. 2023 saw unions representing screenwriters and actors both strike, in part, over concerns GenerativeAI tools could be employed by studios and production companies to replace their work.

Copyright is proving to be the battleground on which this technological and cultural shift is being fought. As of the end of 2023, there is no clear international consensus on what, or to what extent, content produced entirely (or largely) by GenerativeAI tools like Chat-GPT, Midjourney and DALL-E 2 is protectable as a matter of copyright, or at all. This state of flux over what (exactly) is protected—and whether protection even arises in a particular jurisdiction—raises novel complexities and challenges for those looking to acquire or license rights in content. It also raises complications for employers looking to ensure copyright (and other rights) in works produced by their employees properly vests and can be enforced, or even registered.

Rights holders (and prospective rights holders) looking to position themselves ahead of the curve as 2024 rolls in would be well advised to consider the contract terms they use when producing, acquiring, or licensing content that is, or might be, created using GenerativeAI tools, and whether these terms provide sufficient coverage and disclosures to enable monetization and enforcement of rights in that content.

Transparency related to the nature and extent of GenerativeAI's contribution to a finished project is likely to be a key contractual consideration. For example, in the United States, both the Copyright Office and the Courts had taken the position that protection will accrue only to the human author, and only for those portions of a work which have attributable human input rising to the level of "originality—for example, earlier this year the Copyright Office denied copyright existed in visual elements of a graphic novel (Zarya of the Dawn) produced by GenerativeAI, but permitted registration of the written elements and the "selection, coordination, and arrangement" of the written and visual elements attributable to a human author.

Emerging differences between jurisdictions in protection of content created using GenerativeAI likewise highlights the importance of tracking and retaining details about how GenerativeAI was used to create content sought to be protected or monetized. For example, earlier this month, a court in China ruled the nearly 300 prompts input by a human provided sufficient justification to extend copyright protection to the resulting output produced by a GenerativeAI tool (Li Yunkai v Liu Yuanchun). However, in the United States, the U.S. Copyright Office denied any copyright protection for a visual work identified as created solely by GenerativeAI, absent any human involvement. The decision was affirmed on appeal to the US District Court of for the District of Columbia (Thaler v Perlmutter) and is currently subject to a further appeal in the US Court of Appeals, DC Circuit. In Canada, GenerativeAI tools have been listed as the "author" of works recorded on the Canadian Copyright Register, but unlike the US Copyright Office, Canada does not conduct substantive examination of copyright applications, and the validity of such registrations has not yet been tested. There is likely to be future legislation on the subject in Canada—the Federal government recently extended the consultation period on "Consultation on Copyright in the Age of Generative Artificial Intelligence" to January 15, 2024.

With so much change happening so quickly, it is difficult to say, with any degree of certainty, where each jurisdiction will land; or whether there will be any consensus over protection absent a new international agreement or treaty (which would take years to negotiate and implement).

While definite answers on the nature and extent of protection are elusive, there are concrete steps that creators and others that produce, acquire, or license content may consider taking now, to help position themselves to take action while the mosaic of protection for GenerativeAI outputs is coming into focus.

In the absence of legislative guidance, contracts (including service contracts, assignments, and licenses) are a tool that can help create a measure of certainty around treatment of content produced through GenerativeAI, at least as between the parties. The flexibility provided by contracts permits parties to consider the scope and degree of uses of GenerativeAI permitted under the contract, and whether to any representations, warranties, guarantees, disclosures, or indemnities relating to such use should be part of an agreement.

"Future proofing" a business is a never-ending task. In this age of rapid technological change, one of the most important actions rights holders (and prospective rights holders) can take now to position themselves well for 2024 and beyond is to review existing contracts and consider whether the existing terms sufficiently cover the rights to be created, transferred or licensed when Generative AI is used to produce content or complete a task.

The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.