Legal AI

Legal AI and source verification: avoiding persuasive errors

What is the dispute?

AI systems may produce convincing legal answers that are nonetheless wrong. They can confuse judgments, misstate legal tests or provide citations to non-existent sources.

Negotiated solutions

Verification begins with the primary source: read the original judgment or statute, and check the court, date, ECLI, relevant reasoning and factual context.

Proceedings and remedies

Then assess whether the source genuinely supports the proposed conclusion and whether subsequent cases or legislative amendments affect the analysis. A correct case citation is not sufficient.

Legal assessment

Confidentiality and professional duties matter too. Sensitive client material should not be uploaded to external AI services without appropriate prior assessment.

Practical preparation

On LinkedIn, Rob van Biezen discussed receiving two different, incorrect AI answers to a question about the classic Dutch Kribbenbijter judgment. This illustrates why qualified human verification remains essential.

Further reading and primary sources: Rechtspraak.nl · HarveyBloom Advocaten.
AI editorial notice. This text was drafted with AI as general information, and has not yet been individually reviewed or approved by the lawyer. Original Dutch legal sources prevail. Disclaimer.

Relevant practice area · Contact

← All articles