Flood of legal AI tools sparks hunt for fake court rulings

More and more lawyers are using AI chatbots to find case law, meaning previous rulings that can be used as arguments in a case. The problem is that the AI tools sometimes invent rulings that don't exist at all, but look completely real, with years, case numbers and judges' names.

A database that tracks the cases globally has now registered more than 2,000 court cases in which AI has used fake citations or fabricated content. That is a sharp increase from around 200 cases just a year and a half ago.

Costly consequences

Several judges have punished lawyers severely for submitting AI-generated documents without checking them. In June 2026, a judge in Mississippi canceled an entire trial and suspended two lawyers from practicing for two years after both sides had used fake citations. In another case, a lawyer was fined around $110,000, the largest fine of its kind so far.

The reason the AI deceives so convincingly is that legal citations always follow the same template: a name, a year, a case number. The AI model learns the pattern and can therefore easily make up a fabricated ruling that looks exactly like a real one. Since no lawyer knows every previous ruling by heart, the only way to catch the error is to check every single reference manually.

New tools to help

This has created a new niche in the tech industry: programs like Clearbrief and CiteCheck AI, which automatically scan legal documents and flag references that cannot be found in the actual court databases. According to industry information, around 70 percent of the 20 largest law firms in the US already use such tools.

For Danish readers, the point is that the same risk applies if you use a chatbot to research laws, regulations or other professional material: The AI can present fabricated sources with great confidence. It is still the user's own responsibility to check whether what they are told is actually true.

What it means for you

If you use a chatbot yourself to research legal or professional questions, always check the sources yourself; don't trust that they exist just because they look real. If you are a lawyer or work with law, you should consider tools that automatically verify references before documents are sent on. In general, the same applies to all use of AI at work: double-check facts before you act on them.