Lawyers from the US law firm Levidow, Levidow & Oberman were fined for providing information fabricated by ChatGPT. In some cases the details were incorrect and in other cases things were completely thought of by the chatbot.

The incident occurred in a lawsuit against the Colombian airline Avianca. Levidow, Levidow, and Oberman lawyers used ChatGPT to refer to previous similar lawsuits, so The Guardian writes. Lawyers can try to direct cases to the same verdict based on case law, but in this case it turns out that the previous cases are partly or wholly fabricated by ChatGPT.

The law firm was fined nearly €4,600 for the violation. According to the judge in question, Peter Kevin Castle, the lawyers submitted “fake legal opinions with bogus quotes”. Although the court quickly questioned the material, the lawyers then tried to defend their misinformation. The lawyers say they have never provided incorrect information.

The judge in question says there is nothing wrong in principle with the use of AI in legal work, but that the responsibility rests with lawyers: “Technological developments are very common and there is nothing inherently improper in using trusted AI as a tool. But the rules state that lawyers have A supervisory role regarding the authenticity of the submitted material.

It is a known fact that AI language models modify or invent information. Tweakers recently wrote a back story about how these models work and why this could lead to “hallucinating” chatbots.