What Happens When You File a Brief with Fake Citations

In June 2023, a federal judge in the Southern District of New York sanctioned attorney Steven Schwartz $5,000 after he filed a brief containing six fabricated case citations generated by ChatGPT. The cases had plausible-sounding names, realistic volume and page numbers, and even included quotes that appeared to support his arguments. None of them existed.

Schwartz was not the first attorney to file bad citations, but his case became the catalyst for a national conversation about AI in legal practice. Within months, courts across the country began issuing standing orders requiring attorneys to disclose AI use in filings.

The consequences extend beyond fines. In the Eastern District of Pennsylvania, an attorney was sanctioned $2,500 for AI-generated citations in 2024. In Texas, another attorney faced discipline for citing a case that an AI tool fabricated. Each incident follows the same pattern: the attorney trusted the output without checking it against an actual database.

The financial penalties are significant but secondary. The real cost is reputational. Bar associations track sanctions. Opposing counsel finds them. Clients discover them. A single incident of filing fabricated citations can follow an attorney for the rest of their career.

The irony is that verification is straightforward. Every one of these fabricated citations would have been caught by a simple database lookup. The case name does not exist. The volume and page number do not resolve. The reporter series never published it. A tool that checks citations against an authoritative database would have flagged every one.

The lesson is not that AI is dangerous. The lesson is that verification is non-negotiable, regardless of how the citation was generated.