Are AI legal chatbots accurate enough to trust for court filings?

The Appeal of AI Help With Court Paperwork

Court filings — whether a small claims complaint, an answer to an eviction notice, or a motion in a family law case — often involve specific formatting, procedural requirements, and legal terminology that can be intimidating for someone without legal training. AI chatbots have become an appealing option for people trying to draft or understand these documents on their own, since they can explain what a form is asking for and even help generate draft language. The question is whether that help is accurate enough to rely on for something as consequential as an actual filing with a court.

Where Accuracy Breaks Down

The most well-documented accuracy failure involves case citations. General-purpose AI chatbots, when asked to support an argument with legal authority, have in some well-reported instances generated citations to cases that do not actually exist — presenting them with the same confident tone as real citations. This is not a hypothetical risk: a widely covered federal court case involved attorneys who submitted a filing containing fabricated case citations produced by an AI chatbot, resulting in sanctions against the attorneys involved. This incident has become a frequently cited cautionary example across the legal profession about verifying AI-generated legal content before relying on it.

Beyond citations, AI chatbots may also not have current knowledge of a specific court’s local rules, current form versions, filing fees, or formatting requirements, all of which can vary not just by state but sometimes by individual county or courthouse. A chatbot’s general knowledge of “how eviction cases typically work” may not reflect the precise procedural requirements of the specific court where a filing will actually be submitted.

How Courts Are Responding

In response to instances of AI-generated errors appearing in filings, a number of courts have adopted or considered local rules addressing AI use — some requiring attorneys or litigants to disclose when AI tools were used to help prepare a filing, others requiring certification that all citations have been independently verified against a reliable legal database. This is an evolving area, and the specific requirements differ from court to court, making it important to check the rules of the specific court where a document will be filed.

Practical Steps Before Filing

Given these risks, anyone using an AI chatbot to help draft a court filing should treat the output as a first draft rather than a final product. That means independently verifying any cited case law or statute, confirming the current version of any court form is being used, checking the specific court’s local rules (including any AI disclosure requirements), and, where possible, having a court self-help center, legal aid clinic, or attorney review the document before it’s submitted.

Bottom Line

AI legal chatbots are not yet accurate enough to trust blindly for court filings, particularly for legal citations and jurisdiction-specific procedural requirements, and documented cases of fabricated citations leading to sanctions underscore the need for independent verification. Using AI-drafted content as a starting draft rather than a final filing is the safer approach.

Go deeper

• Does Perplexity Cite Its Sources?
• Can AI Chatbots Fully Replace Human Customer Support?
• How Accurate Are General AI Chatbots Compared to Medical-Specific AI Tools?