Editor’s note: This article first appeared in the Multnomah Bar Association’s newletter, Multnomah Lawyer.
Most Americans now use artificial intelligence regularly, and lawyers are no exception. In fact, recent studies by Clio, SurePoint, and 5W show a nearly 80% adoption rate of generative AI throughout firms.
Popular tools among legal professionals range from chatbots such as ChatGPT and Gemini to specialized platforms such as CoCounsel, Clearbrief, and Spellbook. These solutions are designed to help users save time, automate repetitive busywork, and reduce administrative burden. As often occurs with new technology, however, alongside novel benefits come novel risks.
Risk #1: fines for unchecked “facts”
Several recent legal cases have made headlines for involving documents that contain AI hallucinations—information artificial intelligence just makes up.
In Oregon, a federal judge levied a six-figure fine against two attorneys who submitted briefs that referenced over a dozen “non-existent cases” as well as “fabricated quotations falsely attributed to eight legitimate authorities.” Notably, the judge’s opinion omits the hallucinated details in order to prevent the false information from spreading, as AI uses exactly these sorts of documents to generate the output that got the lawyers in trouble.
Perhaps it’s no surprise that fictional precedent is a widespread problem. In a survey of just one month’s worth of proceedings across the United States last summer, Thompson Reuters found 22 “hallucinations and citations of non-existent legal cases,” most of which similarly led to disciplinary actions and sanctions.
Risk #2: attorney-client privilege at stake
Someone who discusses a lawsuit with a conversational AI tool like ChatGPT may forfeit their right to attorney-client privilege. So ruled a New York federal court in early 2026.
The dispute involved a man who asked Claude, Anthropic’s AI assistant, to analyze charges against him and come up with arguments in his favor. The defendant subsequently shared the output with his lawyer, assuming it couldn’t be used against him. Had he waited and asked his lawyer to use Claude instead, the court determined, the material might have been protected from disclosure.
Even so, AI tools typically retain user conversations. Regardless of who talks to a chatbot and when, chat history may be used for AI large language model (LLM) training—and could even be subpoenaed—signaling major privacy concerns.
Risk #3: not-so-great client experiences
AI-related liabilities aren’t limited to the world of case law. Relying too heavily on chatbots and automated assistants for client communication can harm existing relationships and drive away opportunities.
The North Carolina Bar Association’s Practice Management Director, Catherine Reach, writes:
AI can simulate conversation, but it can’t form a bond. And in law, that bond is often what keeps a client engaged, loyal, and confident in the outcome. Delegating that experience to a tool, however advanced, turns a relationship into a transaction.
Consider someone’s emotional state when they call your firm. They may be distressed about the legal problem they’re experiencing. They might be anxious about an upcoming deadline or court date. Chances are they’re looking for comfort, knowledge, and reassurance they can only get from a real person who understands them and can offer meaningful support. AI boosts efficiency, but client relationship-building is inefficient by nature. Retention comes from trust—trust that’s established and maintained over time between people, not a user and a tool.
Using AI to make client service more human, not less
At Ruby, we believe every caller deserves a personal, human experience. That’s why we’ve taken a different approach to integrating AI into our business. Rather than using AI in place of humans, we’ve designed our tools to help our team of receptionists and your practice deliver even better client service:
- Call Nav is our LLM assistant that allows our team to provide detailed answers to virtually any question instantly. While a receptionist handles a call, Call Nav pulls relevant information from a database built for the firm that receptionist is representing. It navigates dozens of frequently asked questions in real time to synthesize key points throughout the conversation.
- AI-generated transcripts provide verbatim records of every call Ruby handles for your business, as well as your voicemail messages.
- Our caller sentiment analysis tool summarizes how callers really feel about your firm. Within each call log, you can see a detailed breakdown of caller sentiments, as well as any phrases the AI has evaluated as negative, so you can determine problem areas and improve quality of life for the people you serve.