A federal appeals court panel rebuked a New York attorney on Friday after he admitted that artificial intelligence tools contributed to citation and quotation errors in briefs filed in an $82 million personal injury case — one of the largest transit injury verdicts in New York City history.

The case involves Luisa Janssen Harger Da Silva, a Brazilian student tourist who lost her left arm and left leg after fainting onto a Brooklyn subway track and being struck by a train in 2016. A jury awarded her $82 million — including $70 million for pain and suffering — finding the Metropolitan Transportation Authority liable for failing to install safety measures that could have prevented her fall.

The MTA is appealing the verdict. But Friday’s arguments at the Second US Circuit Court of Appeals quickly shifted from the merits of the case to the conduct of Da Silva’s attorney.

The AI Problem

Attorney Elliot Shields of Roth & Roth LLP had sent letters to the court ahead of oral arguments acknowledging multiple quotation and citation errors in his brief. US Circuit Judge Maria Araújo Kahn, a Biden appointee, pressed him directly on whether AI was responsible.

“It was not clear to me from reading your letters, but it sounded like some of these errors that occurred in briefing were the result of the use of AI,” Judge Kahn said. “You don’t actually note that in your letters, so I’m asking you whether in fact that was the case.”

Shields confirmed that the errors likely “arose from the use of AI,” calling them “professionally embarrassing and humiliating” in his letters to the court.

Critically, Shields clarified that the AI did not fabricate cases entirely — a problem that has plagued other attorneys. Instead, the issue was that paraphrased summaries were accidentally presented as direct quotes from court opinions.

“We were rushing and we were sloppy,” Shields told the panel. “I think that I transposed notes and put them in quotations.”

Judge Kahn was firm but measured in her response.

“AI can be a wonderful resource,” she said. “But there’s an attorney in the state that I primarily work out of in Connecticut that was recently suspended for precisely this type of issue. The use of AI has its perils.”

US District Judge Rachel Kovner, a Trump appointee sitting with the appellate panel, noted the brief contained “any number of quotes that are not in the cases they’re attributed to.”

The Second Circuit gave Shields one week to file a corrected brief.

Friday’s hearing is the latest in a string of high-profile incidents involving attorneys using AI tools without adequate verification of the output:

  • Earlier this summer, a Washington attorney was sanctioned after using ChatGPT and Claude to prepare filings containing dozens of false citations
  • Last year, a Georgia lawyer was reprimanded for citing AI-generated “hallucinations” in comedian Katt Williams’ assault case
  • In 2023, New York attorney Steven Schwartz made national headlines after his ChatGPT-crafted brief cited non-existent cases, telling a sanctions hearing: “It just never occurred to me that it would be making up cases”

Courts across the country have begun issuing standing orders requiring attorneys to disclose AI use in filings and certify that all citations have been independently verified.

The Underlying Case — and Why It Matters

Beyond the AI controversy, Da Silva’s case raises significant questions about transit safety and the MTA’s duty of care to subway riders.

Da Silva’s attorneys argued that the MTA had more than a decade of internal data showing that exposed subway tracks posed a documented danger to riders — yet the agency failed to install platform screen doors, which are used by major transit systems in cities including London, Tokyo, Singapore, and Paris.

The MTA’s attorney, Max Crema, defended the agency’s position, calling the absence of platform screen doors a reasonable “trade-off” given limited public resources.

“The MTA has studied them in great detail, including by convening a special platform screen door task force,” Crema said. “Ultimately, the MTA, like every other mass transit system in the United States, chose an incremental course, further consideration and pilot project planning.”

The MTA argued it had qualified immunity because the decision not to install platform doors was a deliberate planning decision, not negligent oversight. Crema noted that no other heavy rail system in the United States has adopted platform screen doors.

Judge Kovner questioned whether that immunity would hold if the studies the MTA relied on were “plainly inadequate.”

The $82 Million Verdict

The jury’s $82 million award to Da Silva is considered a potential landmark verdict that could establish significant liability for transit agencies that fail to secure platform edges despite available safety technology.

The breakdown of the award:

  • $70 million for pain and suffering
  • $12 million for additional damages

If the Second Circuit upholds the verdict, it could open the door to similar lawsuits against the MTA and other US transit agencies from riders injured in platform falls — and could accelerate pressure on transit systems to install platform screen doors nationwide.

The three-judge panel — Judges Kahn, Kovner, and US Circuit Judge Reena Raggi, a Bush appointee — did not issue a ruling immediately following Friday’s arguments.

InjuryWatch News will continue tracking this case as the Second Circuit considers the MTA’s appeal of the $82 million verdict.