Artificial intelligence is rapidly transforming the legal and regulatory landscape, but not all tools are delivering on their promise. According to Zeidler Group, the problem lies not in the technology itself, but in how it is being trained — and what it is missing.
The firm has observed a growing trend among legal AI vendors of bolting on additional features to inflate perceived value. New language models, expanded advisory boards, additional industry verticals — the list of enhancements keeps growing. Yet despite this flurry of activity, Zeidler Group argues that something far more fundamental is being overlooked: human feedback grounded in practical, lived experience.
At the heart of the issue is the distinction between blackletter law and how regulation is actually applied in practice. The law itself will always form the baseline for any legal AI tool, but in industries such as investment management, regulatory compliance rarely plays out as neatly as the written rules suggest.
Firms interpret and implement regulations differently depending on their risk appetite, and market practice can span a wide spectrum — from conservative to aggressive — all while remaining broadly within the boundaries of the rules.
This is the gap that static data simply cannot fill. Training a model on legislation and written guidance alone, without accounting for the nuance of real-world application, produces results that fall short of what investment management firms actually need. As Zeidler Group puts it, this is a classic case of garbage in, garbage out.
The firm believes that truly effective legal AI must be treated as a living, dynamic tool — one that is continuously refined through feedback from practitioners who work with the regulations day in, day out. That means curated, contextual and often qualitative data, drawn from the granular, real-time experience of those applying the rules on the ground.
Zeidler Group says it actively solicits feedback from every client and prospective client, using that insight to improve its tools and ensure they reflect how regulations are genuinely applied rather than merely how they are written. The firm argues this approach produces outcomes that far exceed what a generic or even industry-specific AI tool can achieve, provided that tool lacks input from genuine subject matter experts.
For more insights, read the full story here.
Copyright © 2026 FinTech Global
Copyright © 2018 RegTech Analyst
