OTG New York v. Ottogi America: a New Jersey federal court fines a lawyer $3,000 for a reply brief built on AI-hallucinated citations (2025)

The problem

OTG New York, Inc. and Ottogi America, Inc. — rival distributors of the Korean food brand Ottogi — are locked in a contract fight on the U.S. East Coast. As the court recites, “Plaintiff seeks twenty million dollars in damages due to Defendant’s alleged unjust enrichment and breach of contract.” This story is not about who wins that dispute. It is about the fake law one side put in front of a federal judge.

Plaintiff’s counsel, Sukjin Henry Cho, filed a reply brief supporting a Motion to Consolidate. Reviewing it, the court found the cited cases “either (1) did not exist, (2) did not support Plaintiff’s propositions and/or were irrelevant to the issue at hand, or (3) did not contain the supposed quoted authority that Plaintiff asserted.” The court issued an Order to Show Cause why Cho should not be sanctioned under Rule 11.

What was built

There was no legal-tech product here — only unverified generative AI standing in for research. Cho’s own response conceded that “[l]egal research was conducted with the support of generative AI tools and draft summaries, supplemented by targeted LexisNexis database searches.” The court was blunt about the mechanism: “Mr. Cho has confirmed that he indeed used AI, which provided him with hallucinated case law and propositions that he then incorporated into his reply brief without verification.” No specific AI product is named in the order, and LexisNexis appears only as a supplement to — not the source of — the fabricated authority.

The outcome

The court found Cho “violated Rule 11 due to his negligent certification and submission of unverified case law due in part to ‘overreliance on generative [AI] prompts during early-stage research.’” Choosing a sanction “on the lower end of the range” in light of Cho’s prompt admission and remedial steps, “the mitigating factors lead the Court to conclude that a fine of $3,000.00 is sufficient to deter this conduct in the future.”

The order therefore:

• sanctions Cho $3,000.00 under Rule 11(c), payable to the court registry within fourteen days;
• marks the offending reply (ECF No. 42) “WITHDRAWN” and orders it “STRICKEN from the record by the Clerk of the Court”; and
• requires Cho, within fourteen days, to “self-report this Order to the disciplinary authority of the state bars of which he is a member” and to serve his client with a copy.

The $3,000 is the only monetary sanction. The court’s recital that district courts have imposed fines “ranging from $1,000 to up to $6,000” is a survey of other cases, not this one; the $20 million is the damages sought in the underlying dispute.

Two independent newsrooms reported the sanction firsthand. Bloomberg Law: “A New Jersey attorney must pay a federal court $3,000 for misusing artificial intelligence in support of a motion to consolidate cases between feuding ramen noodle distributors.” Daily Voice: “Sukjin Henry Cho, who operates out of Fort Lee, was fined $3,000 by United States Magistrate Judge José R. Almonte for submitting fabricated case law…”

Path to green

This is a court-adjudicated public record; no client-confirmation call applies. The Tier-1 order is byte-tied to the archived RECAP PDF, and two independent reputable newsrooms (Bloomberg Law; Daily Voice) carry the $3,000 + AI firsthand. The remaining human step before a green badge is confirmation that the $3,000 sanction was paid within fourteen days and was not later modified, vacated, or reversed (e.g., on objection to the magistrate judge’s order), and that Cho’s state-bar self-report proceeded.