3 min read · from the paper
What earns lawyers’ trust in AI?
For the lawyers who took part in Zeno’s workshops, the question was already practical: what should AI do in legal work, and what must remain under the lawyer’s control?
What the Dutch Legal Market Wants From AI brings together live polling from approximately 116 Dutch lawyers across two workshop sessions. Natali Petras, Legal Product at Zeno, examines four recurring expectations: capable assistance, output that can be checked, visible reasoning and dependable security.
AI use is becoming routine
Among the 61 respondents who answered the frequency question, 52% used AI every day and 23% used it several times a week. Another 18% used it a couple of times a month, while 7% had tried it but were not using it. None reported never having used it.
Those figures describe the workshop respondents. They show a group already forming views about where AI helps and where it falls short. As use becomes more familiar, the useful question is how well a tool fits the work and supports review.
Assistance with the work, with the lawyer in charge
Research, summarising, drafting, contract review and translation feature prominently in the responses. These are tasks where AI can help prepare material for a lawyer to assess and develop.
The report also describes AI as a sparring partner: a way to explore counterarguments or test a line of reasoning. In each case, the professional remains responsible for judging the result in the context of the matter. A draft still needs review. A suggested argument still needs to be tested.
Verifiable output is central to trust
When asked whether AI can guarantee the accuracy and legal validity of its output, 96% of respondents said no. The finding points to a clear expectation: lawyers want to be able to verify what a tool produces.
That review involves more than checking whether an answer sounds plausible. The lawyer needs to examine the authorities, interpret them in the relevant context and decide whether the conclusion follows. The report places this continuing role for professional judgement at the centre of responsible AI use.
Sources, reasoning and security
Source visibility is another recurring theme. In the workshops, 55% of respondents believed they could see the sources an AI tool used to answer a question. This records their perception, rather than a test of the tools themselves.
The practical requirement is traceability. Lawyers need to follow an answer back to the material supporting it and understand how that material relates to the conclusion. Visible sources give them something concrete to inspect and challenge.
Confidence in security was more divided. The report describes questions about where client data goes, how confidential material is handled and whether inputs are used to train models. For legal teams evaluating a tool, those questions require specific answers about the service and its governance.
About the report
This four-page English report is published by Zeno Research and written by Natali Petras. It combines workshop findings with a discussion of what they mean for legal AI. The polling reflects the participating group, with response counts varying by question, and should be read as directional insight rather than a representative survey of the whole Dutch legal market.
Download the report to read the original discussion and the chart showing how often respondents use AI in their daily work.
Read the full paper
Explore the full discussion and references behind this report.
