As Generative AI (GenAI) continues to weave itself into corporate operations, the debate around its use in legal and compliance work has grown louder. Advocates highlight its efficiency and transformative potential, but experts warn that without human oversight, GenAI risks becoming not just ineffective, but dangerous.
At the centre of this debate is Zeidler Group, a law firm that builds LegalTech solutions powered by GenAI. Drawing on direct experience, the firm argues that in highly regulated industries such as financial services, human input is not just desirable, but required by law.
The limitations of GenAI become particularly clear when applied to the financial sector, where regulators such as the Securities and Exchange Commission and their global counterparts demand sign-off from qualified individuals. Zeidler stresses that success in implementing Legal GenAI tools comes down to two decisive factors: clearly defining a problem to solve, and embedding expert input at every stage.
Recent research backs this up. A study by the MIT NANDA Program found that more than 95% of GenAI projects were generating zero return on investment. This figure is less surprising when one considers how often AI tools are deployed without a clearly defined purpose. GenAI is highly effective when it addresses a genuine business pain point, but without focus it becomes little more than, as Zeidler describes it, “a solution in search of a problem.” In law and compliance, where accuracy is critical, tools need to be tailored to specific challenges rather than broad, generic functions.
Specialisation alone is not enough. Zeidler emphasises that GenAI in legal contexts must be developed and overseen by professionals with real-world experience of regulation and compliance. Simply involving lawyers on paper is insufficient; the tools require practitioners who understand how rules apply in practice. Without that grounding, the risk is akin to an architect designing a bridge without ever having seen a river, it said.
The second cornerstone of effective Legal GenAI is oversight. While much has been made of AI replacing human jobs, the reality is that its strength lies in supercharging existing work rather than replacing accountability. For simple tasks like summarising documents or conducting research, GenAI is a valuable accelerator. But in complex, regulated environments, the final review and approval must remain with humans. In areas like asset management compliance, regulators explicitly require this.
Proposals that suggest using GenAI to oversee other GenAI agents may sound futuristic, but Zeidler dismisses them as impractical. Oversight is not just a governance principle, it is embedded in law. Delegating accountability to an AI system would require rewriting nearly a century of case law in the United States alone, not to mention long-standing frameworks in the UK and Europe.
Looking forward, Zeidler draws comparisons between GenAI today and the internet in the 1990s: transformative but still in need of careful regulation and guidance. The firm argues that Legal GenAI’s role is not to replace professionals but to empower them, bringing speed, efficiency, and scale to processes while keeping responsibility firmly in human hands. Without this foundation of oversight, the promise of GenAI risks becoming a bridge to nowhere.
Read the daily FinTech news
Copyright © 2025 FinTech Global
Copyright © 2018 RegTech Analyst
