Your new co-counsel? Like it or not, it's AI

Three years ago, on Nov. 30, 2022, OpenAI quietly released a free large language model (LLM) chatbot that could seemingly answer any question you had. Chatbots existed before, but they sounded, well, like a chatbot. The brilliance of ChatGPT is that it can process complex information to create new content. Older, rule-based chatbots mostly performed predefined tasks depending on the input. An engineer would have to pre-program the chatbot to identify user queries and provide corresponding pre-programmed responses This presented a crucial limitation because no one could predict every possible query and response in advance. That is why when you try to use the rudimentary chatbot usually present on online shopping platforms, the chatbot often misunderstands your question and ultimately connects you to a live agent. 

ChatGPT has no such limitation. It is built on an artificial neural network and trained on a vast amount of human-generated data to understand patterns from large amounts of information, including text, images, audio and video. When responding to a user request, the chatbot uses the patterns it has learned to predict and create new content. In other words, it's designed to talk like us. And ChatGPT seems to have no question it cannot answer, no problem it cannot solve, not because it has the expertise or is trained to answer those questions or solve those problems, but because it is designed to generate a response based on your prompt, which provides the context.

ChatGPT's success became proof of concept for generative AI: People were not only ready for it, they were enamored by it. And this AI boom has affected every industry, including the legal sector. It is hardly surprising that a profession that constantly demands faster and cheaper work product would turn to a technology promising exactly that. Indeed, traditional legal databases like Thomson Reuters (parent of Westlaw) and RELX (parent of LexisNexis) have worked tirelessly to bring these promised fruits to lawyers through products like Westlaw's CoCounsel and LexisNexis's Lexis+ AI. Brand new platforms like Harvey AI have also begun offering law-specific AI tools that threaten to disrupt the legal technology sector.

Despite these heavily advertised time-saving tools, some lawyers remain skeptical of generative AI, preferring to simply ignore its existence and rely on good old-fashioned boolean searches and human juniors to do grunt work. And headlines support their caution. There has been no shortage of sanctions against lawyers and law firms who include hallucinated case citations, invented quotes, and outright false propositions in their briefs. But reading these widely publicized examples as a general indictment against generative AI absolves lawyers of their responsibilities. As the California Court of Appeals recently held: "Simply stated, no brief, pleading, motion, or any other paper filed in any court should contain any citations--whether provided by generative AI or any other source--that the attorney responsible for submitting the pleading has not personally read and verified." Noland v. Land of the Free, L.P., 114 Cal.App.5th 426, 431 (2025). It is that simple because while technology has no doubt transformed the landscape of legal practice, our ethical duties remain the same.

It is a fundamental duty that a lawyer provides competent representation to clients. Model Rule 1.1 specifies that "Competent representation requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation." Model R. of Pro. Conduct R. 1.1 (2023). Lawyers who rely on AI-generated content without independently verifying the output will likely fall short of this duty. See ABA Formal Opinion 512 at 3-4 ("a lawyer's reliance on, or submission of, a GAI tool's output--without an appropriate degree of independent verification or review of its output--could violate the duty to provide competent representation as required by Model Rule 1.1."). Meanwhile, Model Rule 3.1 prohibits lawyers from knowingly making a false statement of fact or law to a tribunal or failing to correct a false statement of material fact or law previously made. And Model Rules 5.1 and 5.3 specifically require lawyers who have supervisory authority over other lawyers or managerial authority within a firm ("senior lawyers") to make reasonable efforts to ensure that the firm and its lawyers conform to the Rules of Professional Conduct. In certain circumstances, the rules require senior lawyers to be responsible for other lawyers' violations and to ensure that non-lawyers' conduct is compatible with the lawyer's professional obligations. The rules are clear: A lawyer is responsible whether a false citation arises from a junior attorney's human error or an AI tool's hallucination.

The problems do not stop at hallucinations. Recent advancements in hyper-realistic image generation, like Google's Nano Banana Pro, put the authenticity of evidence presented in legal proceedings at risk. What if a client uses these tools to fabricate evidence in their favor? We are operating in a minefield. 

Since we cannot put generative AI back in its box, how do we navigate the vastness ahead? We call for a return to the basics. Every lawyer, aspiring lawyer, and law student should reread their applicable ethics rules and ensure they understand what is required when using AI. Law firms should offer recurring training as technology develops not only to understand how the technology works but also how it intersects with the lawyer's professional responsibilities. Attorneys, young and old, should learn or relearn the foundational skills of legal research and writing without the assistance of AI, both so they learn to think for themselves and so they know what a good work product is to verify what AI has produced. Law firms should also establish and publicize best practices to safeguard against AI errors. For example, to ensure accuracy of case citations, junior lawyers could be required to send PDFs of each opinion cited, and senior lawyers could set aside time to verify the accuracy of the cited cases. Lawyers should keep up to date with the various AI tools and their strengths and weaknesses, and law firms should supply basic AI training to its lawyers.

Your new co-counsel may be AI, but the counsel responsible will still be you.