Skip to main content
Citation Safe

AI Hallucination Sanctions — S.D. New York

47 sanctioned cases on record in S.D. New York where an attorney filed briefs, motions, or other pleadings containing AI-fabricated citations. Every entry below traces to a real, publicly reported court ruling. Data pulled daily from the underlying dockets and reporter databases — last update within 24 hours.

At a glance — S.D. New York

Total cases
47
Date range
2023-06-22 — 2026-08-27
With monetary penalty
5 of 47
Distinct outcomes
20

Outcome breakdown

How courts in S.D. New York have responded when a filing contained fabricated citations.

  • Warning16
  • Admonishment2
  • Bar Referral2
  • No Sanction; Formal Warning Issued2
  • Adverse Costs Order1
  • Adverse Costs Order; Bar Referral1
  • Brief Struck; Default Judgment1
  • Certification Requirement For Future Ai-assisted Filings.1

AI tool implicated

Filings in this jurisdiction most often cite the tool as “Implied” (the ruling describes an AI-generated brief but does not name a product).

  • Implied32
  • Unidentified8
  • ChatGPT2
  • GPT-3.51
  • LexisNexis1
  • LexisNexis AI1
  • NotebookLM; vLex; Paxton AI1
  • Protégé (LexisNexis)1

What this means if you practice in S.D. New York

Across 47 documented sanctions in S.D. New York, the most common judicial response is Warning (16 of 47). Of the 47 cases, 5 carried an explicit monetary penalty in the ruling text.

The rulings in this jurisdiction are consistent with the pattern Citation Safe tracks nationally: when a court identifies a fabricated citation, the sanction almost always follows within days, not weeks. The attorney of record — not the AI — is on the hook. Rule 11 (federal), state analogs, and inherent authority all support monetary sanctions, disgorgement, adverse-costs orders, and bar referrals independent of the tool that produced the citation.

Practical implication: any brief that reaches a docket in S.D. New York needs a citation-existence check before filing. Existence is Layer 1 — the case has to be real. Layer 2 is quote accuracy (any quoted language must actually appear in the opinion). Layer 3 is propositional support (the opinion must actually stand for what you cite it for). Every documented sanction below failed at Layer 1 or Layer 2 — the kind of failure that is trivially catchable before filing and catastrophically visible after.

Sample penalty language from this jurisdiction: “80056 USD”; “2712 USD”; “2500 USD”.

All documented sanctions in S.D. New York

Comparable jurisdictions

Other courts on the Citation Safe sanctions database with a similar volume of AI-hallucination rulings on record.

Weeks with a S.D. New York ruling

Frequently asked questions

How many attorneys have been sanctioned in S.D. New York for AI-fabricated citations?
47 on record as of the latest daily update. The number is a floor, not a ceiling — only rulings where the court explicitly identified the fabricated citations on the docket are counted. Private reprimands, off-record admonishments, and pre-filing catches do not appear.
What is the most common outcome in S.D. New York?
Warning is the most common recorded outcome (16 of 38 rulings with a coded outcome). Courts vary widely in how they label the same underlying disposition — a “warning” in one docket may functionally equal an “admonishment” or “order to show cause” in another.
Which AI tools appear most often in S.D. New York sanctions?
Most rulings (32 of 47) describe the brief as “AI-generated” without naming a specific product. This matches the national pattern: courts rarely require the attorney to identify the tool by name on the record.
Where does this data come from?
Citation Safe ingests published sanctions rulings from public court dockets and reporter databases (federal PACER, state court publishers, appellate reporters). Rulings are extracted, coded for outcome and tool attribution, and republished with attribution to the source court. See /methodology for the full ingest pipeline.
How do I make sure I never appear on this list?
Run every brief through a Layer 1 + Layer 2 citation check before filing. Layer 1 confirms the case exists in a primary source; Layer 2 confirms any quoted language actually appears in the opinion. Both are deterministic checks — not AI second-opinions. Citation Safe offers a free 3-check trial or the Opposition Check $49 flat rate for one-off audits of the other side’s brief.

Verify your own brief before S.D. New York makes it 48 on this list.

3 free checks. No signup. Every citation checked against the primary source — not another AI. $19 unlocks the full annotated report.

Verify a brief free →Check the other side’s brief — $49Free email when a new sanction hits
All jurisdictionsFull sanctions databaseVerify your own brief →

Informational only. Not legal advice. Data derived from public court records; corrections at /contact.