What happened
In September 2026, courts in three US states dealt with the same mistake. In each case, someone without a lawyer used AI to help write their court papers, and the papers pointed to court cases that do not exist.
- Washington: In King v. AT&T, a man suing several companies over his credit report filed papers with several made-up cases. The court was considering a penalty.
- Oregon: In a family law appeal, a person’s brief listed several made-up legal sources. The Court of Appeals fined them.
- Idaho: In Posey v. Bushnell, a man appealing a defamation verdict cited cases that do not exist. The Idaho Supreme Court refused to consider those arguments at all.
Why it went wrong
AI chatbots write in a confident, legal-sounding way. When asked for cases that support an argument, they can make up names and numbers that look real. People without legal training trusted them and did not look the cases up.
Impact
- Real penalties: a fine, and arguments thrown out in a case the person cared about.
- Lost time for courts. The Oregon court said its “resources are diverted” to chasing fake sources.
- A database now counts more than 2,000 court cases worldwide with made-up AI content.
The lesson
- Look up every case you cite. If you can’t find it in a real law library or database, don’t use it.
- Courts treat everyone the same. The Idaho court said all people, with or without a lawyer, must make sure what they file is true.
- A chatbot is not a lawyer. It can help you understand. It can’t be trusted to give you sources.