Article by: Chelsea L. Gulinson & Cober C. Plucker
It’s fun to use generative artificial intelligence to see what your pet might look like as a cartoon. It’s not so fun to be sanctioned for using AI-hallucinated cases in a legal brief.
Like it has with many industries, AI has rapidly transformed the legal profession. AI tools are now widely available to lawyers to assist them with conducting legal research, drafting contracts, reviewing documents, analyzing opposing parties’ filings, and more. These technologies may offer significant efficiency gains, but attorneys and their clients are quickly learning that overreliance on AI may inadvertently expose themselves to serious ethical, financial, and legal consequences.
Perhaps the most widely publicized danger of AI is its tendency to “hallucinate,” i.e., generate information that seems accurate but is false, misleading, or doesn’t exist. That’s because AI generates plausible answers rather than verifying accuracy. An AI-generated legal argument can sound authoritative while actually misstating statutes, inventing precedent, or omitting controlling law.
For lawyers, the risks are obvious: filing false authorities can result in sanctions, reputational damage, and disciplinary complaints. Numerous courts have already sanctioned attorneys for submitting briefs containing erroneous legal authority generated by AI tools. Just a few months ago, a federal court ordered a lawyer to pay over $96,000 in sanctions for filing a summary judgment brief containing non-existent cases and fabricated quotations—even though there was credible evidence that the client, not the lawyer, drafted the brief.[1] Lawyers should be aware of the American Bar Association’s ethics opinions on generative AI as it relates to rules related to competency, informed consent, confidentiality, and fees.[2]
For clients, the consequences are severe. A client may lose a motion, suffer an adverse judgment, incur unnecessary legal expenses, or even have their case dismissed because of inaccurate AI-generated work product.
Lawyers aren’t the only ones at risk of misusing AI. Clients themselves increasingly use AI to do their own research and assess litigation strategies, sometimes in direct challenge to their counsel’s advice. While an AI response might sound promising, it may be inaccurate, incomplete, or based on incorrect or distorted facts, causing clients to overestimate the strength or value of their claims. AI responses also lack strategic considerations that are not obvious from legal standards alone. Even if AI pulls from relevant legal authority in crafting its responses, it may recommend a strategy that looks attractive but that will ultimately damage credibility and make the dispute harder to resolve. Consequently, when attorneys are provided with AI-generated material, they are required to spend substantial time reviewing, correcting, and responding to inaccurate AI conclusions. All of this increases legal fees and distracts from advancing a client’s objectives.
Clients should also be cautious about uploading privileged or confidential information into public AI systems. While disclosing information to a computer program may seem innocuous, it could implicate federal or state privacy and security laws or render the information discoverable in litigation. For example, a federal court recently held that documents generated by a client, who used a public AI system to research legal questions and organize defense theories, were not protected by the attorney-client privilege or shielded by the work-product doctrine.[3]
Overall, clients should be aware that AI does not (at least, not yet!) replace sound advice from a real, skilled lawyer. The legal profession is built upon competence, confidentiality, diligence, independent judgment, and trust. AI may be able to support these objectives, but it can never truly replace them.
For more information, contact Chelsea Gulinson or Cober Plucker at 602-792-3500.
[1] See Opinion and Order (Dkt. 225), Joanne Couvrette et al. v. Mark A. Wisnowvsky et al., U.S. District Court, District of Oregon, Case No. 1:21-cv-00157-CL (March 23, 2026), available at https://websitedc.s3.amazonaws.com/documents/Dkt_225_Opinion_and_Order.pdf.
[2] ABA issues first ethics guidance on a lawyer’s use of AI, American Bar Associates (July 29, 2024), at https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/.
[3] See Edward D. Lanquist et al., Your AI Prompts May Be Discoverable: What Every Client Must Know, Baker Donelson (March 11, 2026), available at https://www.bakerdonelson.com/your-ai-prompts-may-be-discoverable-what-every-client-must-know.
