Lawyers and non-lawyers alike, are increasingly using AI in their divorce cases and ending up in trouble. One Atlanta divorce lawyer who repeatedly cited to phony legal cases in her client’s divorce papers found out that using AI in your divorce can badly harm your case.

Deus Ex Machina

Generative AI is saving lawyers time and money, but if used improperly, it could cost them their livelihoods. Generative AI is a type of AI model that creates new content, such as text, images, or music, based on large amounts of training data.

Examples for text include OpenAI’s ChatGPT, Microsoft’s Copilot, Google’s Gemini, Meta’s Llama, and Anthropic’s Claude, and for images ChatGPT, Gemini, Adobe and Midjourney. A generative AI model that is text-based is called a large language model.

But here’s the thing. When given a question or prompt, AI uses learned patterns to predict the best outputs, such as your motion or brief. Training a generative AI model takes months because of the enormous amounts of data and the neural network’s process of learning and optimization.

Generative AI is impressive but not perfect. AI predictions are just suggestions, and they can be wrong.

Atlanta divorce lawyer, Diana Lynch, is reported to have drafted an order a judge signed in May 2024 that referenced two cases that don’t exist. Lynch’s client filed for divorce in April 2022 and received a divorce decree in July of that year.

In October 2023, the ex-wife asked the county judge to reopen the case and set aside the divorce decree, claiming she had moved to Texas in 2021 and had not been properly served with the complaint. The judge denied the ex-wife’s request in a three-page order in May 2024.

But then the ex-wife’s appealed, and she pointed out the trial judge relied on two fake cases in her order denying her motion, rendering the order void.

Florida Bar and AI Use

I have written about computer and high tech issues in divorce before. Florida was the first state to issue an ethics opinion regarding the use of AI. To ensure client confidentiality, you should confirm that you maintain ownership of any uploaded data, learn how to delete your data, and review the AI provider’s license terms or representations regarding confidentiality.

Free AI models use your questions and uploaded documents to train future models. To maintain client confidentiality, you will need a paid subscription.

U.S. Supreme Court Chief Justice John Roberts warned in his 2023 report on the judiciary that commonly used AI applications can be prone to “hallucinations,” causing lawyers to submit briefs citing fake cases, the Georgia judges said. Hallucinations arise when an AI model generates incorrect or nonsensical information and they are factually inaccurate.

Her

On appeal. a three judge panel on the court of appeals in Georgia found that Ms. Lynch was not deterred when she was accused by her client’s ex-wife of making up case law references. In fact, she filed a response citing 11 cases that were also either fake or irrelevant.

“The irregularities in these filings suggest that they were drafted using generative AI. We are troubled by the citation of bogus cases in the trial court’s order.”

The judges said it might be the first time a Georgia appeals court has confronted the problems that can flow from a lawyer’s apparent use of artificial intelligence that generates content. They said other courts have tackled the issue. They also said Lynch added insult to injury by requesting attorney fees in relation to the ex-wife’s appeal and even used a phony case to support the request.

The judges imposed the maximum penalty on Lynch for her “frivolous” bid for attorney fees. They sent the case back to the DeKalb County judge to reconsider the ex-wife’s request to set aside the divorce decree. Their opinion references a study by Stanford researchers, who found generative AI models, including ChatGPT, “hallucinate” around 75% of the time when answering questions about a court’s core ruling.

The article is here.