Why arbitration may be best route to resolve global AI IP disputes

• Novel issues raised by AI often defy straightforward application of existing legal precedents

• Alternative dispute mechanisms are especially valuable where parties need to resolve disputes quickly to maintain competitive advantage

• Arbitration proceedings provide a secure environment to safeguard valuable trade secrets and other confidential information

Recent developments in generative AI have sent ripples throughout business and society, prompting governments across the world to move to hem in the risks even as they seek to harness the benefits that the technology can bring.

Such initiatives are expected to mitigate the uncertainty that is inherent in the nature and use of AI technology but, at the same time, businesses need to be equipped with the appropriate dispute resolution tools to manage litigation risks as and when they occur.

Indeed, the rapid advancements in AI in recent years have introduced a myriad of cutting-edge legal issues, ranging from IP infringement and licensing violations to various forms of tort-based liability stemming from AI activities. As a result, courts and legislative bodies are still in the relatively early stages of grappling with these advancements.

More specifically, issues facing courts and regulators include:

• use of copyrighted material as training data;

• allegations that AI-generated outputs infringe copyrighted works or violate open source software restrictions;

• use of personal identifiers (name, image and likeness violations);

• training algorithms and generative AI models that may practise patents;

• claims of breaches of data usage agreements for training models;

• discrimination and algorithmic bias; and

• claims for damages potentially resulting from incorrect recommendations or malfunctioning AI systems (eg, autonomous vehicles).

AI-related disputes pose unique challenges and alternative dispute resolution (ADR) avenues (eg, arbitration) are particularly well-suited to resolving such disputes. While other international institutions (eg, the International Chamber of Commerce’s Court of Arbitration (ICC), the American Arbitration Association (AAA), JAMS and the Hong Kong International Arbitration Centre (HKIAC)) have similar rules and practices that govern this field, this article draws primarily from the Singapore International Arbitration Centre’s (SIAC) experience to elaborate.

Factors that favour ADR for AI-related disputes

The unique challenges posed by AI-related disputes are evident from some of the litigation that has already arisen around the world, including IP claims by content owners that the use of their materials to train generative AI systems infringes on copyright.

Among other complaints, there have also been lawsuits in relation to the output of AI tools (eg, where the output is allegedly defamatory of individuals or contains misrepresentations.

In litigating such matters, the complexities that may arise include the determination of:

• liability – given that different players are involved in the development, deployment and use of AI systems, who among the range of actors (eg, AI owners, programmers and users) should be held liable for a wrong committed;

• jurisdiction – given the cross-border nature of AI deployment and usage, whether the alleged acts took place in another jurisdiction such that a national court should not hear the matter (Getty Images et al v Stability AI Ltd [2003] EWHC 3090 (Ch)); and

• evidence – for example, the “black box” nature of AI systems increases the difficulties in determining how an AI system had arrived at a particular output that caused harm.

At the outset, it is important to note that traditional court proceedings can – and do – play an important role in resolving AI-related disputes. However, out-of-court dispute resolution mechanisms may, for the factors highlighted below, assist businesses in resolving various disputes in a more expedient, efficient and effective manner.

Out-of-court dispute resolution mechanisms also have the benefit that they may be more easily enforceable in other jurisdictions. For example, arbitration outcomes, whether monetary or non-monetary, are enforceable worldwide in at least 172 states that are parties to the New York Convention.

Indeed, surveys indicate that international arbitration, used together with other ADR methods (eg, mediation), is the preferred mode of resolving cross-border disputes. Arbitration also continues to be a popular method for commercial dispute resolutions, and the geographic regions covered by arbitration are on the rise.

With the above in mind, we now outline the various factors that account for why ADR is particularly well-suited to resolve AI-related disputes. While these factors are mainly described in relation to arbitration, many facets are likewise applicable toother ADR mechanisms (eg, mediation).

Novelty and complexity of the issues

Perhaps the most elementary factor that favours ADR for resolution of AI disputes is the novelty of the issues raised. These novel issues often defy straightforward application of existing legal precedents. Arbitration allows for custom solutions tailored to novel and complex problems, well-suited for a rapidly developing field such as AI. WIPO, for instance, has served as an ADR forum for copyright disputes surrounding use of training data for AI tools, as well as claims regarding the copyrightability of AI-generated outputs.

Additionally, AI disputes require a deep understanding of the underlying technology. Arbitration offers parties significant autonomy in selecting neutral decision-makers with the necessary technical expertise, unlike generalist courts where judges may lack the requisite knowledge.

This is critical, for example, in disputes involving liability for incorrect AI output or IP ownership of AI systems, to ensure that decisions are well-informed and accurate. SIAC’s panel, for example, has more than 100 arbitrators from over 25 jurisdictions, including a dedicated IP panel of more than 20 thoroughly vetted arbitrators with expertise in IP-related disputes.

Time sensitivity

Moreover, the rapid pace of AI development adds an element of time sensitivity to these disputes. Traditional litigation is typically expensive and time consuming and arbitration may offer a faster, more efficient alternative. The median duration of an arbitration, according to leading arbitration centers such as the SIAC, the ICC and the London Court of International Arbitration (LCIA), range from 11.7 months to 22 months, with an average pendency of 15 months. In contrast, for all commercial cases terminated between 2021 and 2023, the median time to summary judgment was 650 days and the median time to reach trial was 987 days – more than 30 months.

Recognising the urgency that AI-related and IP disputes often carry, arbitral institutions have been proactive in refining their procedures. SIAC can appoint an emergency arbitrator within 24 hours when urgent interim relief is needed, even before the main tribunal is fully constituted. Its rules also allow the tribunal, at the request of a party, to issue an order or award granting an injunction or any other interim relief it deems appropriate. This rapid response capability allows for swift action on critical issues (eg, preventing the unauthorised sale of proprietary technology or maintaining the status quo in a high-stakes licensing agreement).

To further accelerate the resolution process, many tribunals have added early dismissal or early determination procedures to expedite arbitration processes, including SIAC, the HKIAC and the ICC. SIAC’s Early Dismissal mechanism, for instance, permits the tribunal to quickly dispose of claims or defences that are clearly without merit or outside the jurisdiction of the tribunal, reducing time spent on unfounded arguments. SIAC also has expedited or fast-track procedures that can bring disputes to a conclusion within six months. These procedures simplify and shorten the usual timelines, making them ideal for smaller value disputes or where time is of the essence.

Moreover, the no-appeal feature of arbitration, especially to external court systems, ensures that decisions and arbitral awards are final and binding, with limited grounds for challenge on the merits. This eliminates the protracted delays often associated with the appellate process in traditional courts. The AAA has found, for instance, that US district and circuit court cases required more than 12 months longer to resolve by trial than cases adjudicated by arbitration, and more than 21 months longer for those cases that went through appeal.

These alternative dispute mechanisms are especially valuable in the fast-paced AI industry where parties need to resolve disputes quickly to maintain competitive advantage and continue innovation without disruption.

Cost

Cost considerations are also significant. Innovation in cutting-edge tech sectors often includes smaller players and start-ups, which are particularly sensitive to the high cost of legal battles. Arbitration can enable these entities to resolve disputes cost-efficiently, without draining their financial resources and stifling their growth. Tribunals such as SIAC, the ICC and the International Centre for Dispute Resolution (ICDR) calculate fees on an ad valorem basis, meaning they are tied to the amount in dispute, which allows companies to anticipate costs relative to the stakes involved. This approach not only makes arbitration more accessible to smaller entities but also ensures that the process remains proportional and fair.

It is important to note, however, that the greatest cost associated with dispute resolution, whether in court or in arbitration, is legal fees. Regardless of whether the parties decide to pursue litigation or arbitration, such fees will be substantial. Given the streamlined nature of arbitration, as well as the comparatively short duration of a typical arbitration proceeding, however, arbitration will likely yield lower attorneys’ fees than litigation.

Certain organisations (eg, SIAC) enhance cost efficiency by capping arbitrators’ fees and administrative costs, preventing unexpected financial burdens during the proceedings. As fees are tied to the value of the dispute, this cap is particularly beneficial in high-stakes disputes, offering parties financial control without sacrificing the quality of arbitration. A 2018 study underscored SIAC’s cost-effectiveness, finding it to be the most competitive option for both sole-arbitrator and three-arbitrator cases, compared to other major arbitral institutions included in the study.

Confidentiality and privacy

Confidentiality and privacy are paramount in the AI industry, where highly sensitive trade secret training methods and technical implementations are critically important. Arbitration proceedings, with their inherently confidential nature, provide a secure environment to keep such points secret, safeguarding valuable trade secrets and other confidential information. Many arbitral institutions enforce strict confidentiality rules, and parties to the arbitration can request additional restrictions to keep the proceeding private. Under SIAC Rules, for instance, all participants – including parties, arbitrators, SIAC staff, witnesses and experts – are required to maintain the confidentiality of the arbitration. This obligation covers all aspects of the proceedings, including the identities of the parties and the mere fact that the arbitration is occurring.

Privacy is another key advantage. Third parties that are not a party to the arbitration agreement, including the public and the media, cannot attend or play any part in the process, nor will arbitrators publish legal opinions regarding the proceedings. Unlike confidentiality, which varies in degree according to the tribunal and the jurisdiction, privacy applies to all arbitration cases. The privacy feature of arbitration proceeding allows parties to communicate more freely with the tribunal and may help to achieve more efficient processes and better-informed outcomes.

Global reach

Unlike national courts, which are often limited to their own domestic jurisdictions, arbitration offers a global reach that can seamlessly navigate the intricate web of international law and cross-jurisdictional disputes. An arbitral award rendered in one country can be recognised and enforced in a court in any other state that is a signatory to the New York Convention, providing a level of certainty and global enforceability that is unmatched by national courts. This is particularly advantageous for multinationals and entities engaged in cross-border transactions, as it allows for a unified resolution in a single forum, avoiding the pitfalls of conflicting judgments in different jurisdictions.

Arbitration’s remarkable flexibility also extends to the selection of legal frameworks, procedural rules and the venue of proceedings. The choice of arbitration seat – the jurisdiction in which the arbitration is legally situated – can influence both the process and outcome of the arbitration. In the realm of AI and intellectual property, where arbitration can be particularly complex, Singapore has gained prominence due to its progressive IP laws, which recognise IP disputes as arbitrable (of course, this is not to take anything away from the arbitration options available in the United States (eg, the AAA and JAMS)and Europe (eg, the LCIA and WIPO)). This forward-thinking approach aligns well with the global nature of AI and intellectual property, where disputes often involve parties from multiple countries and require efficient resolution mechanisms.

Arbitration also offers a significant advantage in resolving complex cross-border disputes in a single forum, a feature particularly appealing in the context of global IP rights. For example, in cases involving multi-jurisdictional licensing agreements, SIAC arbitration can consolidate multiple proceedings into one, thereby avoiding the risk of inconsistent rulings and reducing both time and costs. This is increasingly the approach in handling global IP licensing programmes, including disputes related to standard essential patents and FRAND terms.

A more adaptable and focused approach

To frame the point in real-world terms, litigation news on AI has been dominated by IP issues. The outcomes of these lawsuits remain uncertain for both copyright owners and AI developers. As these cases proceed to trial, the time and costs will continue to escalate, which may prompt parties to reconsider their litigation strategies and explore ADR mechanisms.

While litigation remains a preferred choice for some parties to set binding precedents and establish points of principle, or for parties to seek injunctive relief given the courts’ level of familiarity with – and enhanced ability to enforce – the measure, the complexity and novelty of AI-related disputes, among the other factors outlined above, demand a more adaptable, focused and efficient approach.

In practice, arbitration has quickly adapted to AI-related issues. Bespoke arbitration rules have been created for AI disputes, including rules regulating the use and disclosure of AI during proceedings (eg, in document discovery to save time and cost).Some of the world’s largest tech companies – from the United States, Asia and Europe – are turning to arbitration to settle their disputes, underscoring its ability to handle the complexities of the tech sector efficiently and effectively.

This article was written for World Trademark Review and was published here on 18 December 2024.