When AI Gives Legal Advice, Confidence Is Not the Same as Correctness
Artificial intelligence is changing the way we work, learn, shop, travel, write, and even think. It can summarize documents, draft letters, organize ideas, and answer questions in seconds. Used carefully, it can be a helpful tool. But when consumers turn to AI for legal advice, one of its greatest strengths—its confident tone—can also become its greatest danger.
I was reminded of this recently when I was contacted by a potential client who believed he had a strong legal malpractice claim. He told me, almost word for word, what Gemini had said. Based on that answer, he was convinced he had a multimillion-dollar case. His reasoning was understandable on the surface: three lawyers were involved, he had read that most lawyers carry $1 million in professional liability insurance, and therefore he assumed the case was worth $3 million.
But that is not how the law works. Insurance coverage is not the same thing as damages. A lawyer may have a $1 million policy, but a client is not entitled to that money simply because something went wrong or because more than one lawyer was involved. The claim still must have value. There must be a breach of duty, causation, and provable damages. In this particular situation, the lawyer may have done something unethical, but the actual economic consequence to the client appeared to be negligible.
I spent about 20 minutes explaining why I could not help him and why the legal claim was not what he believed it to be. I sent him on his way, but I do not think he was convinced. He believed Gemini, not me, despite my 41 years of practice. To borrow a line that Dave Ramsey attributes to his grandmother, “A man convinced against his will is of the same opinion still.”
That is the problem. AI can sound certain even when it is incomplete, misleading, or simply wrong. It may give an answer that seems logical, but law is rarely decided by logic alone. Legal rights depend on statutes, deadlines, facts, local court rules, exceptions, documents, contracts, damages, burdens of proof, and the judgment of people trained to evaluate those details.
Another recent matter showed the problem from a different angle. I was hired to defend a lawsuit brought by a contractor against a subcontractor. The complaint had been prepared by the contractor’s office manager and filed in her name as if she was the company. It was labeled as an action for replevin, but I doubt the office manager even understood what replevin meant. The complaint was a mess. I called the office manager, who did not answer, and left a message explaining that she appeared to be engaging in the unauthorized practice of law and recommended that the contractor immediately retain an attorney to assist with the claim. In Indiana, a person who engages in the business of practicing law without being admitted by the Indiana Supreme Court commits a Class B misdemeanor. Fortunately, the contractor did retain counsel, and I now have an attorney on the other side with whom I can work to resolve the dispute. Before that, the matter was at a standstill.
Whether that complaint was drafted with AI assistance, an internet form, or simple guesswork, the lesson is the same: legal words are not magic words. Replevin, malpractice, negligence, damages, insurance limits, jurisdiction, and standing all have specific meanings. Using the right-looking label does not mean the claim is legally sound. In fact, a little legal vocabulary without legal judgment can make a problem harder to resolve, not easier.
This same pattern is now appearing across the country. Courts have sanctioned lawyers for filing briefs that included AI-generated cases that did not exist. Public trackers and legal researchers have identified hundreds of court orders and sanction matters involving fabricated citations, false quotations, or inaccurate AI-generated legal claims. The American Bar Association has warned lawyers that the use of generative AI does not change their duties of competence, confidentiality, communication, candor, supervision, and reasonable fees. The National Center for State Courts has likewise warned that AI can produce fabricated case names, statutes, legal standards, court procedures, and factual assertions that appear authentic but are not.
If trained lawyers can be misled by AI, ordinary consumers should be especially careful. The danger is not only that AI makes mistakes. The danger is that it makes mistakes assuredly. It gives the user a polished explanation, often with confident wording, and sometimes with citations or legal terms that sound official. The result can be a false sense of certainty.
That false certainty can cause real harm. Someone may miss a deadline because an AI tool misunderstood the statute of limitations. Someone may file the wrong document in the wrong court. Someone may believe they have a claim worth millions when they do not. Someone may disclose private information into a public AI system without understanding where that information goes. Someone may reject a reasonable settlement because a chatbot suggested a much larger number without understanding the evidence or the law.
None of this means consumers should never use AI. It can be useful for organizing questions, learning vocabulary, preparing a timeline, or understanding what documents may be relevant. A person meeting with a lawyer might use AI to help list the facts they remember or generate questions to ask during the consultation. But AI should not be treated as your lawyer, your judge, your insurance adjuster, or your guarantee of recovery.
A safer approach is simple. Use AI as a starting point, not an ending point. Do not rely on it to determine whether you have a case. Do not assume that a general answer found online applies in Indiana. Do not mistake available insurance for collectible damages. Do not upload confidential documents unless you understand the privacy risks. And when the matter involves a legal right, a deadline, a contract, a lawsuit, an injury, an estate, a business, or a family dispute, talk to a qualified lawyer before acting.
The law is built on details. AI is built on patterns. Patterns can be helpful, but details decide cases. The best legal advice comes from applying the law to the actual facts, in the correct jurisdiction, with professional judgment and accountability. AI may be able to point you toward a question. It should not be the final answer.