Artificial Intelligence is finding its way into all aspects of our lives, including how we communicate with one another. However, open-ended AI based on the World Wide Web is not yet a reliable source of legal precedent. In practice, I see many individuals attempt to use open-source AI to obtain insights for legal analysis or case law. I see even more reports and articles about attorneys being sanctioned for using open-ended AI. Case law research using popular AI tools like ChatGPT or Gemini can result in a phenomenon referred to in the legal community as “hallucinations.” In other words, open-ended AI searches can often generate cases that do not actually exist. In addition, sometimes the search will return a reference to an actual case, but that case does not actually establish the legal precedent claimed. Judicial officers really do take the time to read cases, and in the event their time is wasted on a hallucination or a misrepresentation of case law, the attorney will pay the price. In certain jurisdictions, some courts have even adopted standardized rules setting forth the sanction for an attorney’s use of AI.
There are some exceptions to the rule. There are paid subscriptions to closed-end AI search engines that attorneys may subscribe to (Lexis Nexis, Thomson Reuters West), which do use AI-type searches, but those are confined to databases and banks of actual case law, which are more reliable. These subscriptions are not free and are generally contracted for law firm resources.
Non-attorneys should be aware that generating a legal memorandum or brief using open-ended AI is not a substitute for an attorney’s due diligence and should not be used in court. Open-ended AI searches for case law cannot be relied upon in lieu of legal advice or legal research. While AI continues to grow in presence, it should not be used as a primary source for obtaining case law or legal analysis.