Incident·WS-0008·Jun 22, 2023·in the wild·Misinformation & fabrication

Lawyers sanctioned for ChatGPT's invented court cases

Levidow, Levidow & Oberman (Schwartz, LoDuca) / ChatGPT

29/100

WS-2 Moderate

Harm score 20, signal score 39. Autonomy: relied on (people trusted bad output). How the scores work.

Harm 1/5

damage actually done

Reach 1/5

who was affected

Autonomy 1/4

how much the AI did on its own

Novelty 3/5

how new the behavior was

Permanence 1/5

how hard it is to undo

In a personal-injury suit against Avianca, plaintiff's lawyers filed a brief citing judicial opinions that ChatGPT had invented, then defended them after opposing counsel could not find them. Judge P. Kevin Castel (S.D.N.Y.) fined the two lawyers and their firm and ordered letters to each judge falsely named as author.

Damage

$5,000 sanction; six non-existent opinions attributed to real judges

The capability behind it

GPT-4 scores in the top 10% on a simulated bar examMar 14, 2023 · OpenAI · reliable for a defined task: partial

Sources

Abajournal, second source. Date is the sanctions order. Fabricated citations were filed in early 2023 (an April 25 affirmation kept citing them) and became public in May 2023.

Factor ratings and the score are this index’s judgments by a published formula; facts, dates and figures come from the linked sources. Where a lawsuit is pending, the summary reports allegations, not findings.