As a newly-admitted attorney who has undergone both undergraduate school and law school during the rise of artificial intelligence, I have personally seen peers use artificial intelligence for legal research. Although both Westlaw and LexisNexis have both published AI search models within the past two years, both models only provide suggestive responsive answers and precaution users to research the generated information. In practice, I have also seen a rise of pro se litigants using AI tools to write legal briefs, conduct research on case law and statutes, and even write emails.

While AI can be a useful search tool, pro se litigants and attorneys alike should be aware of the following concerns:

AI Usage in Family Law Matters

According to the Manhattan divorce lawyer, the rise of free, publicly available artificial intelligence models such as ChatGPT over the past few years has posed a moral and legal question to attorneys and pro se litigants who seek to use AI search engines to answer legal questions. Although AI can be a useful tool for both searching the internet and populating quick and loquacious responses, the current versions that are publicly available cannot accurately answer complex legal questions. Please be advised that the consequences for using AI-generated legal briefs, or even using artificial intelligence to conduct research on a case, can be detrimental to your case. Attorneys and pro se litigants alike should take extreme precaution in using artificial intelligence to research family law issues, draft legal documents, or seek any kind of logistical legal advice.

The Georgia Court of Appeals recently fined an Atlantic-based divorce attorney after a judge ruled she had used artificial intelligence in a proposed order. The attorney was found to have cited multiple artificially-generated cases in her filing, and the court sanctioned her $2,500, the maximum penalty the law allows in that jurisdiction. The Court thereafter referred the case back to the lower court for reconsideration. The reasoning behind this decision was that the attorney had essentially deprived her adversary, a pro se litigant, from properly responding to her motion because the cases she had cited were fictitious. Essentially, a court can sanction both attorneys and pro se litigants for using false information, including AI-generated case names.

Issues specific to family law

Family law is often fact-specific; courts often consider emotional or relational context when making judicial determinations in topics such as child custody, child support, spousal support, and property division. AI cannot, as of yet, authentically replicate the human judgment required to passionately advocate for emotional considerations. Family law requires human judgment, empathy, and competent understanding of the current applicable law. As such, Manhattan divorce lawyer says that users of AI search engines and generator models should be extremely skeptical and take great precaution in relying solely on AI for guidance regarding family law issues.

Privacy concerns

Most publicly available AI models use cloud-based or third-party servers and, as a consequence, any use of such models may risk exposure of private information. ChatGPT, the most commonly used free AI model, often has leaks of private information because of its lax privacy settings. Searches made on ChatGPT can be published online and found through a simple Google search. Although such searches often do not publish the name of the searcher, any private confidential information included in the search is always at risk for publication and leaks.

Further, some AI models are trained to memorize and regurgitate information given by its users. Although most AI models are coded to filter out information that would be considered personal such as passwords and full names, it is still highly possible that the model will retain and could accidentally regurgitate sensitive facts from your case to other users.

Lastly, like with other

Bias 

AI models are simple search engines that gather information from the internet at large and formulate one cohesive answer as a summary of all the information it is able to compile. As such, answers in any topic formulated by AI will often incorporate information published to the internet decades ago, information from public forums with civilian users like Reddit, and opinion-based think pieces from blogs. Users should be cautious that AI models may propagate an answer based on the personal opinions and biases of such websites rather than on case law, statutes, and legal precedent.

Irrelevant and outdated information

In an area as sensitive and constantly evolving as family law, AI models cannot accurately distinguish relevant and current information from irrelevant and outdated information.

Vitally, AI models often cannot distinguish legal articles talking about New York divorce law, for example, from an article talking about New Mexico divorce law. The search engine will search for key terms and ignore terms that do not apply. A user could hypothetically ask an AI model to summarize how to serve a document in New York and receive a response that is a compilation of information pertaining to New York, New Jersey, and/or New Mexico.

Furthermore, AI models will often source their responsive information from a wide variety of sources throughout the internet. This is inherently problematic to New York divorce law because it is a constantly evolving field where new groundbreaking judicial decisions pass through the Appellate Division each month. As a result, an AI model pulling a response to a question about fault-based grounds for divorce may source its response in part from an overturned case from 1999, an outdated blog article from 2015, and a semi-accurate Facebook comment from 2023. The combination of all its sources will often result in a half-correct or even blatantly false response.

The bottom line

Although AI can be useful for simple functions like correcting grammar and spelling, formatting written material, and even coming up with out-of-the-box strategies, its current models should not be used in any part of a family law case. If you do use artificial intelligence to research your case or draft a legal paper, please disclose that information to an attorney and seek further advice regarding your matter. AI responses, as detailed above, are often inaccurate, and the usage of false AI information can lead to monetary sanctions or even the dismissal of your matter. Even if you receive a response from an AI search engine in your family law case, it is vital to do the requisite legal research or contact an attorney to verify the information is accurate, current, and supported by statute or case law.

In short, I precaution anyone using AI search engines to conduct legal research to at the very least fact check each piece of information. However, in the context of New York family law, I highly suggest bringing your research to an attorney for verification that the information is still applicable under current case law. Without access to legal search tools like Westlaw or LexisNexis or the requisite legal knowledge, it can be dangerous to entrust your sensitive case in the hands of a tool only meant to regurgitate information from a wide array of sources from across the internet.