When Lawyers Trust AI Blindly: Court Sanctions and How This Firm Works Differently

AI is not the problem. Unverified AI is.

Since 2023, courts in New York and around the world have sanctioned lawyers and litigants for filing briefs built on artificial intelligence output that no one checked — fabricated cases, invented quotations, citations to opinions that do not exist. Ronald S. Cook, P.C. has followed this developing body of law closely, because the firm uses AI in its own practice and believes clients deserve to know exactly how. Attorney Ronald S. Cook holds a J.D., dual LL.M. degrees, and an MBA, and has practiced in New York for over two decades — long enough to know that no tool replaces a lawyer reading the law.

Call toll-free: (888) 275-2620. Available 24/7.

The Numbers: A Court Problem That Barely Existed Three Years Ago

Researchers now track AI hallucinations in court filings as a distinct category of judicial decision. The most comprehensive public collection, the AI Hallucination Cases Database maintained by researcher Damien Charlotin, contained 1,871 records as of August 11, 2026, spanning courts in the United States, Canada, Australia, the United Kingdom, and beyond. Reported decisions have increased sharply: comparing the same January-through-early-August period year over year, 2026 records are up roughly 241% over 2025.

The United States accounts for roughly 69% of all records — and New York courts, federal and state, account for more than 110 of them. This is not a distant problem. It is happening in the Southern District of New York, the Appellate Division, and county Supreme Courts where this firm practices.

Self-Represented Litigants Generate the Incidents. Lawyers Face the Consequences.

Most records involve people representing themselves — about 59% of the database — which is understandable: a person without legal training asks a chatbot for help and files what it produces. Lawyers appear in about 38% of records. But severity flips the picture entirely: of the 133 records in which a court imposed a professional sanction, 129 — 97% — involve licensed lawyers. Courts warn pro se litigants. They sanction attorneys.

The money is real, though usually not ruinous — the professional damage is worse. Among quantifiable U.S. dollar consequences in the database, the median is about $2,000 ($2,500 for lawyers), roughly 70% fall at $5,000 or below, and about 88% at $10,000 or below. The largest recorded figure — $110,204 — includes attorney’s fees and costs, not a fine alone. What no median captures: struck briefs, disqualification from cases, bar referrals in 76 records, public reprimands, and court-ordered letters telling clients their own lawyer filed fabricated law. New York courts have been consistent about where the duty sits: the failure is not using AI — it is filing its output without verifying it against real authority.

It Happened in New York — Repeatedly

Mata v. Avianca, Inc. (S.D.N.Y. 2023) is where the era began: a brief citing multiple nonexistent cases with fabricated quotations, continued reliance on them even after opposing counsel raised the alarm, $5,000 in sanctions, and court-ordered letters to the client and to the judges falsely credited with the invented opinions.

Cassata v. Michael Macrina Architect, P.C. (Sup. Ct., Suffolk County, Jan. 27, 2026) — in this firm’s home county — counsel filed a brief containing unverified authorities. Result: $10,000 in monetary consequences and the brief struck. The lesson the court drew is the lesson of this entire page: submissions must be verified against reputable, authoritative sources.

Landberg v. City of New York (App. Div., 2d Dept., June 23, 2026) — nonexistent New York cases, including a fabricated Court of Appeals authority, in an appellate filing. Result: $10,500 in sanctions against counsel and the firm, imposed by the appellate court that hears appeals from Long Island’s trial courts.

LiveVideo.AI Corp. v. Redstone (S.D.N.Y., June 9, 2026) — hallucinated cases and quotations, with remnants of AI drafting prompts left visible in filed documents. Result: $80,056 in adverse costs and a referral to bar authorities.

And beyond New York, courts have gone further: in Johnson v. Dunn (N.D. Ala. 2025), fabricated ChatGPT citations produced a public reprimand, disqualification of counsel from the case, and a bar referral — a court refusing to let unverified AI become an acceptable shortcut.

Even Legal Research Tools Are Not a Free Pass

Honesty requires saying this plainly: the database includes sanctions involving professional legal research platforms, not just consumer chatbots. General-purpose tools dominate — ChatGPT is the most frequently identified tool by a wide margin — but records also name AI features of Westlaw, LexisNexis, and vLex-related products. In Brooks & Brooks Inc. v. Patel (D. Minn. 2025), filings prepared with a vLex Fastcase tool were stricken, with a $5,000 sanction and a bar referral. In Flycatcher v. Affable Avenue, a brief prepared with several tools contained 14 fake citations among 60 authorities.

The pattern across every one of these cases is identical: no lawyer checked the output against the underlying authority before filing. Better tools reduce the risk. They do not eliminate the lawyer’s duty. Any firm that tells you its software makes verification unnecessary is telling you something this body of case law contradicts.

How This Firm Uses AI: Grounded in Real Law, Then Verified by a Lawyer

Ronald S. Cook, P.C. uses Clio Work with Vincent AI, grounded in Clio Library, which is powered by vLex — a legal research collection of actual statutes, regulations, court opinions, dockets, and secondary sources. When the firm uses AI for research, its answers are backed by cited legal authorities drawn from that library, not by a chatbot’s reconstruction of the internet. “Real AI for lawyers” means exactly this: AI connected to authoritative legal sources and used inside a lawyer-controlled verification process — not generic AI output treated as legal authority.

The product helps. The workflow protects the client. AI can accelerate the research; it does not get the last word. Before any material legal proposition or citation reaches a court, it is checked against the underlying authority — the actual opinion, the actual statute. Clio Work and Vincent AI assist the research process; they do not replace attorney judgment, source review, or citation verification. That verification step is precisely what was missing in every sanctions case described above, and it is the step this firm does not skip.

You can read more about the firm’s broader approach to technology on our AI in Legal Services page, our article on AI adoption and legal strategy, and our data security page. If you are facing a dispute where the quality of legal work will matter, our litigation attorneys handle matters across all 62 New York counties.

Contact Us

Call toll-free: (888) 275-2620. Available 24/7.

Our law firm has over 3,000 client testimonials across Google, BBB, Trustpilot, and other platforms. View our verified client reviews.

Contact the law firm

Client reviews and ratings reflect individual experiences and do not guarantee a similar outcome. Prior results do not guarantee future results.

Last reviewed by Attorney Ronald S. Cook — August 2026

This page is for informational purposes only and does not constitute legal advice. The sanctions cases described are matters involving other lawyers and other firms, cited from the public record.