The Dangers of Using Artificial Intelligence (“AI”) in Legal Documents or Research

With the rise of Artificial Intelligence (“AI”) and its ability to create and write documents, the allure of using AI to draft

legally significant documents and perform legal research is hard to resist. These AI tools can be incredibly useful to stimulate ideas, draft outlines, create well-written emails, and outline complex ideas in an easily digestible format. However, a line must be drawn when it comes to using AI tools to draft legally binding or significant documents.

AI is not infallible, and it often experiences “hallucinations”, which is what happens when AI produces false or fabricated information or law that is presented as fact, but is actually false or simply does not exist. These “hallucinations” can happen when AI tries to cite to binding case law or interpret state and federal statutes. Attorneys have fallen for the allure of using AI to draft motions in court cases. These attorneys end up being heavily reprimanded by the Court, and some attorneys’ licenses have been revoked based on the severity of their use of AI to draft a document.

Another critical aspect of using AI tools to draft legal documents or perform legal research is the risk to confidentiality. If you enter confidential or sensitive information into a public AI tool, you risk making that information publically accessible and lose any protection from discovery. Internet browsing history can be discoverable information in the course of litigation. If you are entering private information into AI tools to draft or perform legal research, that information is now possibly discoverable. As the saying goes, “once it is on the internet, it is there forever”. This differs greatly from disclosing confidential information to your legal counsel, which is protected under the attorney/client privilege. Your confidential information stays confidential when you tell it to your legal counsel; the same can not be said for what you tell AI.

AI tools can certainly be used to assist in drafting initial outlines for documents or doing initial research into a legal question, but it is not advisable to rely on AI for final, legally-binding documents or basing actions on the legal research provided by AI. There is a huge risk that the final legal document or legal research articled by AI is incorrect because of hallucinations, lack of context, and even failure to account for specific jurisdictional requirements. Seeking legal counsel, even simply to review an AI-generated document for accuracy and legal compliance, can prevent a lot of pain and suffering down the road.

Written by Of Counsel Attorney Rachel Awalt