- Court
- New York Supreme Court, Appellate Division, Second Department
- Case
- Matter of Julien v. Arthur
- Docket
- 2024-03307
- Filed
- May 27, 2026
- Slip Op
- 2026 NY Slip Op 03308
- Citation
- 2026 NY Slip Op 03308 (N.Y. App. Div. 2d Dep’t 2026)
Background
Kevin A. Julien and Danielle M. Arthur are the parents of a child born in June 2020. The father filed a petition for parental access, and the mother filed for sole legal and residential custody. The proceedings were complicated by allegations that the father failed to return the child after scheduled visits and assaulted the child’s maternal relatives during custody exchanges. After a consolidated hearing, the Family Court, Kings County granted the mother sole legal and residential custody with limited supervised parental access for the father. The father appealed pro se.
During the appeal, the Appellate Division issued an order to show cause directing the father to explain why sanctions should not be imposed pursuant to 22 NYCRR 130-1.1 for filing a brief containing fabricated legal authority. The court identified multiple citations in the father’s appellate brief to cases that did not exist — hallmarks of content generated by artificial intelligence tools without human verification. The sanctions motion was referred to the panel hearing the appeal for determination.
Holding
The Appellate Division, Second Department affirmed the Family Court’s custody determination and, in a significant ruling of first impression, held that the unverified use of generative artificial intelligence (GenAI) to draft an appellate brief containing fabricated legal citations constitutes frivolous conduct warranting the imposition of sanctions under 22 NYCRR 130-1.1, even when the offending party is a pro se litigant.
On the custody merits, the court found that the Family Court’s determination was supported by a sound and substantial basis in the record. The evidence established that the father engaged in a pattern of failing to return the child after visits and committed acts of violence during custody exchanges. The court deferred to the Family Court’s credibility assessments, and the totality of the circumstances supported awarding sole custody to the mother with supervised access for the father.
On the sanctions issue, the court addressed what it characterized as “an emerging issue with broad and significant implications” — the use of GenAI in drafting legal briefs. The court found that the father’s brief contained citations to nonexistent cases, a well-documented consequence of using AI language models to generate legal content without verifying the citations. The court held that submitting fabricated authorities to a court — regardless of whether the fabrication was intentional or the result of unverified AI output — constitutes frivolous conduct. The court imposed a sanction against the father.
Takeaways
This decision establishes important precedent at the appellate level in New York regarding the use of generative AI in legal proceedings. The court’s holding makes clear that attorneys and pro se litigants alike bear responsibility for verifying every citation and legal assertion in their submissions, regardless of how those submissions were drafted. The use of AI tools does not diminish a filer’s obligation to ensure the accuracy of the brief, and the submission of fabricated citations — whether knowingly or through negligent reliance on AI output — will be treated as frivolous conduct subject to sanctions.
The decision is notable for extending sanctions to a pro se litigant. While courts often afford pro se parties greater latitude in procedural and substantive matters, the Second Department drew a clear line: fabricated legal authorities are sanctionable regardless of the filer’s status. The court recognized that the integrity of the judicial process depends on the reliability of the authorities cited by the parties, and that AI-generated fabrications undermine that integrity just as deliberately invented citations would.
Practitioners who use AI tools for legal research or drafting must implement verification procedures to ensure that all citations correspond to actual cases and that the cited cases stand for the propositions attributed to them. Simply copying AI-generated output into a brief without independent verification is not an adequate standard of practice.
Why It Matters
This is one of the most significant appellate decisions to date addressing the use of generative AI in legal practice. As AI tools become increasingly prevalent in both law offices and among self-represented litigants, courts across the country are grappling with how to address the submission of AI-generated fabrications. The Second Department’s decision sends an unambiguous signal: the courts of New York will not tolerate fabricated citations, and the use of AI is not a defense to sanctions for frivolous conduct. This ruling will likely influence other courts considering similar issues and may prompt the adoption of formal rules or guidelines governing the use of AI in legal submissions. For the legal profession broadly, the case underscores that while AI can be a valuable tool for legal research and drafting, it must be used responsibly, with human oversight and verification as non-negotiable requirements.