The use of AI, and in particular generative AI, has expanded steadily in recent years. What began as a peripheral tool is now becoming embedded in everyday legal practice. Legal research platforms increasingly deploy natural language processing to distil large volumes of case law, while contract review tools rely on machine learning to identify and flag potentially problematic clauses. As a result, AI is not only reshaping how legal services are delivered, but also how legal risk is identified and managed.
A less frequently examined consequence of this shift, however, is the emergence of AI-related disputes. As the technology becomes more deeply integrated into business operations, the likelihood of disagreement inevitably increases. Last year’s ruling in Getty Images v Stability AI marked a significant early development in this area, representing one of the first UK judgments to grapple directly with the intersection of copyright and trademark law and the creation and use of generative AI systems. More broadly, it stands among the earliest major AI disputes to reach the courts.
Future AI litigation is unlikely to be confined to the IP-based claims seen in Getty. Instead, it is likely to encompass a broader spectrum of disputes, including issues relating to data sourcing, model training, system deployment, and ongoing oversight. In this way, the continued integration of AI into legal and commercial practice will not only transform how lawyers work, but also expand the scope and complexity of the disputes they are required to resolve.
Jurisdictional challenges
Before looking at the different types of AI related disputes that might arise, it is worth addressing the broad issue of jurisdiction.
In last year’s Getty v Stability AI case, Stability AI was able to avoid liability on the basis that its model had been trained outside the UK. This outcome is likely to encourage AI developers to structure their activities across borders, including training models in jurisdictions with more permissive copyright regimes to mitigate legal exposure.
Jurisdictional complexity is not limited to IP claims. Non-IP causes of action will often need to be brought where the relevant harm occurs, yet with AI systems this can be difficult to determine. Development, training, and deployment may all take place in different locations, making it unclear where legally significant acts have occurred. This fragmentation creates scope for “forum shopping”, with companies selecting favourable jurisdictions for different stages of the AI lifecycle. For claimants, this may necessitate parallel proceedings in multiple countries, targeting both the place of model training and the location where harm ultimately materialises.
Licensing disputes
The continued expansion of data scraping to train AI models creates a dilemma for rights holders and publishers. On the one hand they might seek to protect their data and rights by pursuing multi-jurisdictional copyright litigation. On the other hand, they could seek to negotiate pre-training licence agreements with AI developers, without which access to content and data scraping will be blocked.
This second strategy is already emerging. Major publishers, music catalogues, and image libraries are establishing AI training licensing programmes and demanding that AI companies pay for training rights. However, when licensing negotiations break down there is a risk of legal action being taken – particularly with regards to claims that publishers and rights holders could have for unlawful website scraping by AI developers.
The result is an escalating tension between rights holder and AI companies which will drive a surge in licensing disputes in 2026 and beyond.
Breach of Contract and/or Negligence
Claims involving breach of contract and professional negligence, based on AI deployment, will also likely start to appear in the not-too-distant future.
The fundamental challenge that contracting parties will need to deal with, lies in the inherent unpredictability of AI systems, particularly those employing machine learning algorithms. Unlike conventional software that follows predetermined rules, AI systems can evolve and produce outcomes that diverge from initial specifications or reasonable expectations. When an AI tool or service fails to deliver promised functionality, underperforms against benchmarked standards, or produces erroneous outputs, questions of contractual breach become significantly more nuanced. For instance, was the breach caused by the AI developers flawed data and/or inadequate training methodology or did the supplier utilising AI to deliver a service, breach its obligations to exercise sufficient human oversight over the outputs. The answers to these questions, amongst others, will rarely be straightforward and will create complex liability issues that will fall to be assessed against the contractually defined responsibilities and obligations, underpinning the party’s relationship.
Where contractual obligations are uncertain, claimants might be forced to turn to negligence, as an alternative cause of action. Professional negligence claims will be equally complex and could arise from allegations based on inadequate due diligence on AI model limitations, failure to implement appropriate testing and validation protocols or insufficient consideration of bias and fairness issues.
Conclusion
As organisations re-evaluate their digital strategies and set increasingly ambitious targets for AI adoption, an increase in AI disputes is inevitable. Questions around data provenance, model governance, contractual allocation of risk, and regulatory compliance are no longer theoretical concerns but live issues that will, in many cases, give rise to disputes.
At the same time, the technical and operational complexity of AI systems makes these disputes more difficult to anticipate and manage. Responsibility may be distributed across multiple actors, including developers, deployers, data providers, and third-party vendors, each operating under different contractual frameworks and, often, in different jurisdictions. It is therefore essential that organisations engage proactively with partners, consultants, lawyers, and other specialists to navigate this evolving landscape.
