When a Chatbot Rewrites the Prenup
A Toronto family lawyer recently drafted a prenuptial agreement exactly as his client asked, then received an email listing 13 more changes, nearly as long as the agreement itself. According to a recent Wall Street Journal article by Dalvin Brown, some of the edits were so granular they did not read like something a person would write. The lawyer knew his client had consulted an AI chatbot, and he knew that explaining where the chatbot went wrong would mean billing the client for extra time. It is a vivid example of what can go wrong with AI divorce advice.
Family law attorneys across the country are describing versions of this story. People feed their side of a dispute into a chatbot, get an answer that sounds authoritative, and bring it to their lawyer as settled fact.
AI can be a useful tool. But in a divorce or custody case, the gap between a confident answer and a correct one can cost real money and real credibility. Here is what we want anyone relying on AI divorce advice to know.
Why People Turn to AI First
It is easy to see the appeal. Divorce is stressful, legal fees are a real worry, and a chatbot is available at 2 a.m. for little or no cost. People use it to understand unfamiliar terms, summarize documents, draft messages to an ex, and get a rough sense of what to expect.
Used that way, AI can help people show up to a first meeting better prepared. The trouble starts when a chatbot’s answer is treated as legal advice.
Where AI Divorce Advice Goes Wrong
A chatbot can give you the answer you hoped to see, and that answer may not hold up in front of a judge. Four problems come up again and again.
- It only hears your side. A chatbot works from what you type. It cannot test your account against the other party’s, and it tends to build on whatever framing you give it.
- The law is local. Massachusetts has its own alimony statute and child support guidelines, and Rhode Island’s Family Court has its own rules and procedures. Generic answers often blur one state’s rules with another’s, or with no state’s at all.
- It can invent things. Chatbots sometimes produce case citations and legal rules that do not exist. Courts have been tracking a growing number of family law rulings involving AI-generated errors, and filings built on bad citations can lead to delays, extra costs, and sanctions.
- Judges notice. Documents and arguments that read as machine-generated can cost you credibility with the court, and credibility matters when a judge is deciding contested issues.
There is also a cost problem. As the Toronto example shows, a lawyer who has to untangle a chatbot’s suggestions is spending billable time on it. Starting from a correct baseline is usually cheaper than correcting a wrong one.
What You Tell a Chatbot Is Not Protected
A conversation with your lawyer is protected by attorney-client privilege. A conversation with a chatbot is not. In United States v. Heppner, a federal criminal case in New York, a judge ruled earlier this year that a defendant’s chats with an AI tool about his legal situation were not privileged, even though he had reportedly used material he received from his own attorneys. The reasoning is simple: an AI tool is not your lawyer, so what you type into it is not a confidential attorney-client communication.
Heppner was not a divorce case, but the lesson carries straight over. Divorce involves broad disclosure of finances, and anything that is not privileged may be requested by the other side and used as evidence against you.
The takeaway is to treat a general-purpose chatbot like a conversation with a stranger. Do not put any of the following into one:
- Anything your attorney has told you, or documents your attorney has sent you
- Financial details such as income, account balances, assets, debts, or account numbers
- Your custody, support, or settlement strategy, including what you would accept or where you are flexible
- Candid descriptions of events that could hurt your case if the other side read them
The same caution applies to documents. Many documents in a divorce contain sensitive financial or privileged information, so we generally advise against running your documents through an AI tool at all.
If you have already shared information like this, tell your attorney so they can plan for it.
Using AI Sensibly During a Divorce
We are not telling anyone to avoid AI altogether. We are asking you to use it for the right jobs.
Reasonable uses
- Learning what a term like “equitable division” or “parenting plan” generally means
- Organizing a timeline of events or a list of questions for your attorney
Better left to your attorney
- Deciding what you are entitled to under Massachusetts or Rhode Island law
- Drafting or revising agreements and court filings
- Choosing a strategy on custody, support, or property
- Responding to the other side during an active case
If a chatbot tells you something surprising, bring it to your attorney as a question, not a conclusion. A good lawyer will tell you whether it applies to your case, and that conversation is protected.
Talk to a Person Before You Act on a Chatbot’s Answer
If you have been researching your situation with AI and want to know what actually applies to you, we can help. Hera Law Group offers a free 15-minute initial consultation by phone. Bring your questions, including the ones a chatbot raised, and we will tell you what the law says and what it means for your case.
Our attorneys serve clients from offices in Maynard, Milton, Cape Cod, Nantucket, and Worcester, Massachusetts, and Cumberland, Rhode Island. Visit heralawgroup.com to schedule your consultation.
This post is for general information and is not legal advice. Reading it does not create an attorney-client relationship.
Source: Dalvin Brown, “People Are Using AI for Divorce Advice and It’s Driving Their Lawyers Crazy,” The Wall Street Journal.
