The owner of an online wellness company was paying hefty spousal support. Taking far less out of his business, he asked a chatbot whether a judge would cut his payments. The bot told him he had an easy win. It never asked to see his finances or his spending. What the chatbot called an easy win took two law firms and nearly two years to sort out.
Family lawyers say this is now common. In a spring survey of 133 lawyers at one national family-law firm, 80% said clients had sent them correspondence, draft filings or strategies they believed were generated by AI. The reasons are human. Money is tight when a marriage ends. Hiring a lawyer takes time. A chatbot answers right away, even at midnight. The problem is that it tends to agree with whoever is typing. Feed it one side of a fight and you get the answer you want, even if a judge won't.
When the judge notices
Judges are losing patience. A researcher who tracks AI errors in court has counted 148 family-law rulings where judges flagged fake citations or other apparent AI mistakes, more than 100 of them this year. A Georgia pastor paid $19.99 a month for a chatbot instead of a $7,000 lawyer. It pointed him to a custody rule he misread, and he ended up apologizing to the judge. A Florida man representing himself filed at least 145 motions and sent an appeals court an 863-page appendix. The court ordered its clerk to reject his future filings unless a lawyer signs them. In the wellness-company case, the chatbot had simply accepted the man's story about his finances. A forensic accountant later found what it left out. He likely overpaid support the whole time, and everything he paid his first lawyer was, one attorney said, money down the drain.
The lawyers are careful to say AI isn't useless. It's just that family law, where two sides disagree on the facts, is a bad place for a tool that takes one side at its word.
Our read
This is a Family and Money story, and it brushes Estate planning too. The lesson isn't that AI is bad. It's that a free tool that tells you what you want to hear is dangerous exactly where the stakes are personal and the facts are contested. Divorce is the sharpest example. But the same trap waits in any money decision where you're emotional and the tool hears one side. A beneficiary fight. An inheritance. A business split. An aggressive tax idea you found online.
Here's the useful line to draw. Use these tools to learn the vocabulary and the questions, not to make the decision. A chatbot is fine for understanding what a QDRO is or what spousal support generally considers. It is not fine for whether a judge will rule your way, because it can't see the other side's evidence and it wants to please you. Here's the check this week for anyone going through a big family-money change. Before you act on any advice, free or paid, write down the one fact that would blow up the plan if it's true, then go find out whether it is. In the story, that fact was the husband's real spending. Naming it is what a good lawyer, or a good advisor, does that a chatbot won't. If you're in the middle of one of these transitions, the paperwork, the beneficiary forms, the account splits, is worth a careful human read. That's a conversation we're glad to have.