The global disputes and investigations landscape is being transformed by the evolution in artificial intelligence (AI) and the emergence of new regulatory frameworks. These developments are reshaping how legal professionals approach complex matters, from early-stage investigations to courtroom proceedings.

At its heart is the growing use of AI for speed and efficiency, with tools being used to streamline traditionally labor-intensive processes, enable faster case assessments and improve risk identification.

AI-assisted investigations are becoming increasingly sophisticated, capable of analyzing vast datasets to uncover patterns, anomalies, and connections that would be difficult or impossible to detect manually. This capability is particularly valuable in cross-border disputes and regulatory inquiries, where speed and accuracy are critical. Courts and tribunals are also beginning to adopt AI technologies to support judicial functions, making it imperative to ensure transparency, accountability, and fairness.

AI’s impact is also clear in the area of disclosure. AI is revolutionizing how legal teams manage document review, using natural language processing and machine learning to prioritize relevant materials, reduce costs, and mitigate human error.

At the same time, legislation is evolving around these advancements as regulators aim to balance innovation with ethical and legal safeguards. Against this backdrop, a clear understanding of AI’s capabilities and limitations is vital not just for businesses and legal firms but also for judges, arbitrators and regulators.

This article explores how AI and emerging legal frameworks are reshaping the disputes and investigations ecosystem. It examines the opportunities and challenges presented by the following key topics:

• Efficiency in legal process 
• AI-assisted investigations 
• Judicial adoption 
• AI in disclosure

Efficiency in legal process 

The legal process here specifically relates to disputes and not general legal operations. Disputes cover a multitude of mechanisms including litigation, arbitration, mediation and adjudication. But all of them follow a similar process from a workflow perspective, if not from a legal one. The typical stages of a dispute include pre-dispute analysis, pleadings, disclosure (which is beyond the scope of this article), pre-trial/hearing, the trial/ hearing itself, judgement, appeals and resolution. AI can impact all of these different stages, but the areas beyond disclosure and document review where it can make a significant difference are in case preparation, legal research, the hearing itself and predictive analysis.

In case preparation, AI tools can greatly empower the legal team by generating detailed case timelines and summaries from correspondence and case files, and helping legal teams structure factual narratives efficiently. It can also help identify gaps, inconsistencies, and weaknesses within the documents relevant to the case, allowing these to be addressed and considered as part of the preparation of the case.

Many might have expected AI to undermine the case for using well-established software solutions, but so far, this has not been borne out. Firms are still concerned about the security of client data, integration to existing workflows and the dangers of hallucination. Over the past year we have seen software vendors addressing these concerns as they add AI functionality to their solutions.

The risks of using public AI tools are becoming well understood, especially around hallucination, where the AI response generated sounds superficially plausible but is not grounded in underlying data. For lawyers, using public AI tools without thoroughly verifying the results themselves can have serious personal consequences. In the recent UK case of R (Frederick Ayinde) v London Borough of Haringey, a legal team was required to pay wasted costs after facing criticism for submitting non-existent cases into pleadings, while in Alabama the court disqualified three attorneys for submitting fake citations.

AI’s power to assist in the response to the other side, be it in terms of disclosure or arguments, is linked to the benefits derived from looking at one’s own case to identify gaps, inconsistencies, and weaknesses.

During the hearing itself, most people are familiar with courtroom technologies that allow for documents to be brought up in real time for the participants to review, along with the transcription of the hearing, which is also available to remote participants. AI could make a significant impact here, assisting with - in effect - real time fact-checking and document-checking as someone is giving evidence. The system could interpret the words, analyze what is being sent, compare that to the rest of the case documentation, and then present suggested findings in the form of follow-up questions, or pinpoint errors. Such an AI assistant could enhance the capabilities of counsel during a hearing.

Predictive analysis is also an area where AI can provide great insights. Without AI, parties to the litigation are relying on the expertise of their lawyers and counsel, as well as the financial experts for views on quantum. AI brings the ability to scan all prior cases, compare the factors of those cases with your case, and make predictions based on the AI analysis, which can be used in conjunction with the lawyers’ advice. This can also be refined once the judge/tribunal panel is known, to limit the analysis to only include cases where they sat. This should give parties better insights into their case, although it is worth exercising caution as every case has unique nuances that may result in different decisions.

AI-assisted investigations 

In the realm of financial investigations, where the effective analysis of structured and unstructured data is key, proper deployment of AI can have a profound impact on the investigator’s ability to rapidly detect complex fraud schemes, revealing instances of collusion that perpetrators employ to circumvent established controls. Specifically, AI systems excel in analyzing vast datasets across systems, recognising patterns, and detecting anomalies that may elude traditional approaches.

For example, in cases of collusion, where individuals work together to manipulate or override existing controls, the advanced algorithms and machine learning capabilities of AI enable the identification of subtle and coordinated efforts to deceive systems and exploit vulnerabilities. Furthermore, AI’s ability to adapt and learn from emerging patterns of fraudulent behavior by employing unsupervised learning methods, such as clustering and anomaly detection, contributes to its effectiveness in detecting new and evolving types of fraud across large and complex data sets even across systems. GenAI is only building on these strengths, but when it comes to investigations, it is important that all types of AI are considered as they can drive different benefits.

The overwhelming benefit of AI is in speed and scope. It allows users to parse a greater amount of material in a shorter timeframe, thereby making more informed and timely decisions for clients. These gains can be dramatic, collapsing the time taken for review by as much as 90%, and in the process, reducing costs by 60% or more, largely by removing the need for human review of thousands of documents.

When it comes to investigations, AI is being used in several key areas to drive these benefits:

1. Combining structured and unstructured data 

As mentioned above, AI excels in analyzing vast datasets across systems, therefore it is important to an investigation to bring in different sets of data, both structured (e.g. financial records) and unstructured (e.g. emails) to one environment to allow them to be analyzed in conjunction with each other. Furthermore, Natural Language Processing (NLP) is used to analyze data such as emails, reports, chat logs, and social media content to identify potential misconduct, manipulation or insider trading. It also has the capacity to incorporate external data sources, including open-source intelligence (OSINT) databases, to enhance the contextualization of transactions and relationships, which may not be readily apparent.

2. Document review and analysis 

GenAI excels at reviewing large volumes of documents, identifying relevant materials, and flagging potential issues speedily. Combined with other AI techniques such as sentiment analysis, these can provide investigators the capability to meticulously evaluate sentiment surrounding specific actions or communications. By leveraging AI, legal professionals can distil thousands of data points in a concise and effective way, allowing them to quickly identify key documents and to understand their position faster than ever. This is crucial in the early stages of an investigation when clarity is of the essence.

3. Case strategy and reporting 

GenAI can assist in building case strategies by organising key documents, identifying “helpful” and “harmful” documents, creating chronologies, and even drafting initial reports. For instance, it can analyze documents related to a specific individual and generate a first draft of a report explaining their involvement in a case. This functionality streamlines the preparation process and allows legal teams to focus on higher value tasks.

4. Short-form communication analysis 

Traditional algorithms often struggle with short-form communications and messages, due to the lack of extensive text. GenAI is far more effective at analyzing these types of communications, including emojis in context, making it a valuable tool in investigations where such messages are critical evidence.

5. Proactive compliance and fraud prevention 

Companies have been using AI for a long time to try and detect and prevent fraud and other economic crimes, especially in financial transactions, where AI excels in anomaly detection by learning from historical data, recognizing patterns, and adapting to evolving trends. GenAI also has a role to play here where companies can use it proactively for compliance purposes. For example, AI can scan internal communications for suspicious language or conduct periodic compliance reviews by sampling data from key custodians. Another focus for corporate users is the pre-emptive detection of attempted fraud - for instance, in the insurance markets. These approaches, previously time consuming and costly, are now more feasible due to the efficiency and cost-effectiveness of AI tools.

6. Judicial adoption 

For AI to be truly revolutionary in the dispute environment, the judiciary and equivalent institutions in arbitrations need to be not just onboard with the use of such tools, but also knowledgeable about them. Much of it will come down to ensuring judges and arbitrators are well-informed about AI tools, techniques and emerging technologies, so they can recognize both the power and the perils of AI. By doing so, they will be empowered to make the necessary judgements to drive efficiencies, and to use AI themselves.

The UK government has introduced an “AI Action Plan for Justice” (Justice AI Unit) which sets out a strategic focus on three areas: strengthen foundations, embed AI across the justice system, and invest in people and partners. From this plan, there a number of key goals the UK government has outlined to adopt AI cautiously but strategically, focusing on enhancing efficiency, maintaining fairness, and upholding public trust. These include:

• A phased approach to implement models across the judiciary system once they have been though comprehensive testing in a controlled environment. These should both look to drive efficiencies in administration as well as providing ‘new’ tools to the public to use to aid their access to justice.
• Issuing guidance to ensure that AI is used responsibly – avoiding erroneous results through hallucinations, for example. However, the focus is on empowering human decision making, not replacing it. Therefore the drive will be to put tools in users’ hands to enable better and swifter decisions.
• Transparency, fairness and human oversight are seen as critical, ensuring that there is no embedded bias, that people are aware of not only the use of AI but of how it is being used, and that ultimately there is a human making the decisions.
• Training and awareness will be, as it must, a key element, to drive adoption and familiarity with AI, as well as to ensure stakeholders understand AI’s capabilities and limitations.
• Implementation of appropriate safeguards to ensure continued faith and trust in the legal system, as well as the security and privacy of sensitive and valuable data.

AI in disclosure 

While there are various facets to the use of AI in disclosure, one area where we are seeing the most progress and success is in the impact of large language models (LLMs).

In the e-discovery process, GenAI models use pretrained LLMs as a reference framework for completing elements of discovery including document identification and review. Using these LLMs, GenAI is able to deliver accurate results ‘out of the box,’ saving significant amounts of time during the process. Perhaps more surprising is that these new systems also deliver better quality results than a human review, as measured by elusion and overturn rates. The largest effect has been in relevance with instances of “Not Relevant” proving to be 100% accurate and “Relevant” scores exceeding 90% accuracy.

GenAI also allows users to perform a number of tasks beyond document classification and sorting. For instance, GenAI can perform conceptual searches allowing discovery experts to use the technology for fact finding by inserting a prompt into the system. The GenAI model uses LLM technology to swiftly answer the question using natural language answers, references and example documents. Moreover, the natural language processing capabilities of GenAI also enable it to perform these searches and analyze sentiment across multiple languages simultaneously.

What is key throughout the use of GenAI is the development of the prompts that are used to generate the results and the quality control analysis being performed on it. It is critical that the prompts are developed in a professional manner and not used like a casual online search. Time must be spent on setting the background, setting the personas, detailing the type of output needed, the specific technical terms used and so on, to generate the best results. In our experience this often takes a couple of iterations but is greatly enhanced through collaboration with a prompt engineering expert.

The quality control aspect is equally important. By now the industry is familiar with the perils of AI hallucinations in general use, and also with instances of GenAI being used to produce statements to courts, where referenced cases do not exist, for example. Therefore, performing appropriate checks is vital. Here, the more traditional analytical tools available through technology-assisted review (TAR) can be used to verify the results and provide a basis for the use of GenAI in courts as TAR is recognised and approved by courts.

We have seen the correct use of GenAI drive huge advantages to clients both in terms of cost and time savings during the disclosure review process, but also in the initial fact-finding stage of an investigation whereby a true understanding of the situation can be achieved swiftly, allowing for appropriate legal strategies to be adopted. When dealing with regulators, this can be hugely advantageous.

Conclusion 

AI, and GenAI in particular, will continue to revolutionize the disputes and Investigations landscape for the foreseeable future. When considering its use, be it for efficiency in legal process, AI-assisted investigations, judicial adoption or disclosure, it is important to develop a clear understanding of all the tools and techniques available so they can be deployed in an effective manner whilst also managing the associated risks.

* This report was first published in Mealey’s Litigation Report: Artificial Intelligence.

[Editor’s Note: Phil Beckett is a Managing Director and Leader of Alvarez & Marsal’s European and Middle East Disputes and Investigations practice in London. Any commentary or opinions do not reflect the opinions of Alvarez & Marsal or LexisNexis®, Mealey Publications™. Copyright © 2025 by Phil Beckett. Responses are welcome.]