Should I use ChatGPT or another AI to analyze my bankruptcy?
Use AI to get smarter; don't ask it to be your lawyer. Chatbots are genuinely good at explaining concepts, helping you organize facts, and generating questions to ask. They're unreliable at legal conclusions: they only know the facts you thought to mention, may miss your state's rules or current law, and can state wrong things confidently. And never paste identifying or account information into a public chatbot. Arrive prepared; let a lawyer supply the judgment.
In this answer
What AI is genuinely good for
- Understanding concepts. “Explain a cramdown like I’m tired” is a perfect chatbot request, and asking follow-ups until something clicks is what the tools do best.
- Organizing your facts. Turning a shoebox of worries into the six lists from the know-your-numbers answer — categorizing debts, structuring a timeline of what happened — is clerical work AI handles well, with the privacy rule below.
- Generating questions. “What should someone with a repaid family loan and an unfiled tax year ask a bankruptcy attorney?” produces a genuinely useful list to bring with you.
Where it fails, specifically
A chatbot’s analysis of your case inherits four problems at once. It knows only the facts you volunteered — and non-lawyers reliably omit the facts that matter most (the transfer, the decree clause, the tax year), because you can’t flag what you don’t know is relevant. It may not know your state’s exemptions, your district’s practices, or the law as of this year — and this Library’s own research found respected websites carrying stale figures, a problem AI training data compounds. It can be confidently wrong, citing rules and cases that don’t exist. And it renders conclusions without accountability: when it’s wrong about whether your house is safe, nothing happens to it.
Why the printout backfires
Handing a lawyer a long AI analysis feels like preparation; it usually functions as anti-preparation. The lawyer can’t see what you asked, what facts went in, or what assumptions the machine made — so before giving advice, they must now un-analyze the analysis, checking conclusions against facts they haven’t gathered yet. You’ve added a layer to remove. Bring the inputs instead: your facts, your documents, your questions. Those make the hour better. A verdict to defend makes it worse. (And if AI told you to do something — move money, transfer a title, pay someone back — do nothing until a lawyer hears it. The what-gets-cases-in-trouble answer is a museum of confident, catastrophic pre-filing moves.)
The privacy rule
Never put into a public chatbot what you wouldn’t put on a postcard: Social Security numbers, account numbers, logins, or documents containing them. Describe your situation in categories and rough numbers — “about $40,000 in cards, a car worth $12,000 owing $9,000” — which is all a concept-level conversation needs anyway. This applies to every tool, this site’s included: GoBK doesn’t ask you to upload sensitive documents, and nothing here needs your name.
The bottom line
The useful division of labor: technology to understand, organize, and prepare; a licensed human to judge, decide, and be accountable. Use the machines to become the best-prepared client in your lawyer’s week — then let the lawyer do the one thing the machines can’t: be responsible for the answer.
Sources
This is general information, not legal advice. The right answer for you depends on details a website cannot see, and rules vary by state and by court.
More in Before you hire a lawyer or back to the Library.